Structural Persistence in Language Models: Priming as a Window into Abstract Language Representations
Open accessThe DOI matches its Crossref record; 18 of the 18 references it cites were found in a registry or an index.
Abstract
We investigate the extent to which modern neural language models are susceptible to structural priming, the phenomenon whereby the structure of a sentence makes the same structure more probable in a follow-up sentence. We explore how priming can be used to study the potential of these models to learn abstract structural information, which is a prerequisite for good performance on tasks that require natural language understanding skills. We introduce a novel metric and release Prime-LM, a large corpus where we control for various linguistic factors that interact with priming strength. We find that Transformer models indeed show evidence of structural priming, but also that the generalizations they learned are to some extent modulated by semantic information. Our experiments also show that the representations acquired by the models may not only encode abstract sequential structure but involve certain level of hierarchical syntactic information. More generally, our study shows that the priming paradigm is a useful, additional tool for gaining insights into the capacities of language models and opens the door to future priming-based investigations that probe the model’s internal states.1
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372
Transformers: State-of-the-Art Natural Language Processing
Wolf, T., Debut, L., Sanh, V., Chaumond, J., Delangue, C., Moi, A., Cistac, P., Rault, T., Louf, R., Funtowicz, M., et al. (2020) · Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations · doi:10.18653/v1/2020.emnlp-demos.6 · Cited by 8,741
Chomsky, N. (1957) · doi:10.1515/9783112316009 · Cited by 7,885
DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
Sanh, V., Debut, L., Chaumond, J., & Wolf, T. (2019) · arXiv · doi:10.48550/arxiv.1910.01108 · Cited by 4,653
Syntactic persistence in language production
Bock, J. K. (1986) · Cognitive Psychology · doi:10.1016/0010-0285(86)90004-6 · Cited by 1,911 · cited 12×
“Examples from the core and specialised conditions can be found in Table 1.”
BackgroundMethodInfluentialA primer in BERTology: What we know about how BERT works
Rogers, A., Kovaleva, O., & Rumshisky, A. (2020) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00349 · Cited by 1,552 · cited 2×
“The rationale behind this paradigm is that if speakers are sensitive to sentence structure independently from sentence content, then it is reasonable to assume that such structural information is an integral part of the representations built during processing.”
BackgroundMethodBERT Rediscovers the Classical NLP Pipeline
Tenney, I., Das, D., & Pavlick, E. (2019) · Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · doi:10.18653/v1/p19-1452 · Cited by 1,307
DIALOGPT : Large-Scale Generative Pre-training for Conversational Response Generation
Zhang, Y., Sun, S., Galley, M., Chen, Y.-C., Brockett, C., Gao, X., Gao, J., Liu, J., & Dolan, B. (2020) · Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations · doi:10.18653/v1/2020.acl-demos.30 · Cited by 1,076
Structural priming: A critical review.
Pickering, M. J., & Ferreira, V. S. (2008) · Psychological Bulletin · doi:10.1037/0033-2909.134.3.427 · Cited by 1,047 · cited 2×
“The log probability is computed as the sum of token log probabilities of the LM: For example, the Priming Effect of the example in the introduction would be computed as follows: (where P PO , P DO , T PO , T DO denote sentences 1a, 1b, 2a, 2b).”
BackgroundMethod
Research works printed but not linked: 73
- Alishahi, 2019, Analyzing and interpreting neural networks for NLP: A report on the first BlackboxNLP Workshop, Natural Language Engineering, 25, 543, 10.1017/S135132491900024X · DOI
- Baroni, 2022, On the proper role of linguistically-oriented deep net analysis in linguistic theorizing
- Bernolet, 2010, Does verb bias modulate syntactic priming?, Cognition, 114, 455, 10.1016/j.cognition.2009.11.005 · DOI · Paper page
- Black, 2021, GPT-Neo: Large scale autoregressive language modeling with mesh-tensorflow, 10.18653/v1/2022.bigscience-1.9 · DOI · Paper page
- Bock, 1989, Closed-class immanence in sentence production, Cognition, 31, 163, 10.1016/0010-0277(89)90022-X · DOI · Paper page
- Bock, 2007, Persistent structural priming from language comprehension to language production, Cognition, 104, 437, 10.1016/j.cognition.2006.07.003 · DOI · Paper page
- Boleda, 2020, Distributional semantics and linguistic theory, Annual Review of Linguistics, 6, 213, 10.1146/annurev-linguistics-011619-030303 · DOI · Paper page
- Branigan, 1999, Syntactic priming in written production: Evidence for rapid decay, Psychonomic Bulletin & Review, 6, 635, 10.3758/BF03212972 · DOI · Paper page
- Branigan, 2006, The role of local and global syntactic structure in language production: Evidence from syntactic priming, Language and Cognitive Processes, 21, 974, 10.1080/016909600824609 · DOI · Paper page
- Bresnan, 2007, Predicting the dative alternation, 69
- Chang, 2000, Structural priming as implicit learning: A comparison of models of sentence production, Journal of Psycholinguistic Research, 29, 217, 10.1023/A:1005101313330 · DOI · Paper page
- Cleland, 2003, The use of lexical and syntactic information in language production: Evidence from the priming of noun-phrase structure, Journal of Memory and Language, 49, 214, 10.1016/S0749-596X(03)00060-3 · DOI · Paper page
- Cochran, 1977, Sampling Techniques
- Davies, 2009, The 385+ million word Corpus of Contemporary American English (1990–2008+): Design, architecture, and linguistic insights, International Journal of Corpus Linguistics, 14, 159, 10.1075/ijcl.14.2.02dav · DOI · Paper page
- Davis, 2020, Discourse structure interacts with reference but not syntax in neural language models, 396, 10.18653/v1/2020.conll-1.32 · DOI
- Dubey, 2008, A probabilistic corpus-based model of syntactic parallelism, Cognition, 109, 326, 10.1016/j.cognition.2008.09.006 · DOI
- Fares, 2017, Word vectors, reuse, and replicability: Towards a community repository of large-text resources, 271
- Fine, 2013, Evidence for implicit learning in syntactic comprehension, Cognitive Science, 37, 578, 10.1111/cogs.12022 · DOI · Paper page
- Foster, 2007, Avoiding repetition in generated text, 33, 10.3115/1610163.1610170 · DOI
- Zihao, 2021, A theoretical analysis of the repetition problem in text generation, 12848
- Futrell, 2019, Neural language models as psycholinguistic subjects: Representations of syntactic state, 32, 10.18653/v1/N19-1004 · DOI · Paper page
- Goldwater, 2011, Structural priming as structure-mapping: Children use analogies from previous utterances to guide sentence production, Cognitive Science, 35, 156, 10.1111/j.1551-6709.2010.01150.x · DOI
- Gries, 2005, Syntactic priming: A corpus-based approach, Journal of Psycholinguistic Research, 34, 365, 10.1007/s10936-005-6139-3 · DOI · Paper page
- Gulordava, 2018, Colorless green recurrent networks dream hierarchically, 1195, 10.18653/v1/n18-1108 · DOI · Paper page
- Hessel, 2021, How effective is BERT without word ordering? Implications for language understanding and data privacy, 204, 10.18653/v1/2021.acl-short.27 · DOI
- Jennifer, 2020, A systematic assessment of syntactic generalization in neural language models, pages 1725–pages 1744
- Hupkes, 2018, Visualisation and ‘diagnostic classifiers’ reveal how recurrent and recursive neural networks process hierarchical structure, Journal of Artificial Intelligence Research, 61, 907, 10.1613/jair.1.11196 · DOI · Paper page
- Ivanova, 2017, Do you what I say? People reconstruct the syntax of anomalous utterances, Language, Cognition and Neuroscience, 32, 175, 10.1080/23273798.2016.1236976 · DOI
- Ivanova, 2012, The comprehension of anomalous sentences: Evidence from structural priming, Cognition, 122, 193, 10.1016/j.cognition.2011.10.013 · DOI
- Florian Jaeger, 2013, Alignment as a consequence of expectation adaptation: Syntactic priming is affected by the prime’s prediction error given both prior and recent experience, Cognition, 127, 57, 10.1016/j.cognition.2012.10.013 · DOI · Paper page
- Jeffreys, 1961, Theory of Probability
- Józefowicz, 2016, Exploring the limits of language modeling, CoRR
- Jumelet, 2020, diagNNose: A library for neural activation analysis, pages 342–pages 350, 10.18653/v1/2020.blackboxnlp-1.32 · DOI
- Jumelet, 2021, Language models use monotonicity to assess NPI licensing, 4958, 10.18653/v1/2021.findings-acl.439 · DOI
- Kaschak, 2011, Structural priming as implicit learning: Cumulative priming effects and individual differences, Psychonomic Bulletin & Review, 18, 1133, 10.3758/s13423-011-0157-y · DOI
- Kass, 1995, Bayes factors, Journal of the American Statistical Association, 90, 773, 10.1080/01621459.1995.10476572 · DOI · Paper page
- Kassner, 2020, Negated and misprimed probes for pretrained language models: Birds can talk, but cannot fly, 7811, 10.18653/v1/2020.acl-main.698 · DOI · Paper page
- Kodner, 2020, Overestimation of syntactic representation in neural language models, pages 1757–pages 1762, 10.18653/v1/2020.acl-main.160 · DOI
- Lakretz, 2021, Mechanisms for handling nested dependencies in neural-network language models and humans, Cognition, 213, 10.1016/j.cognition.2021.104699 · DOI · Paper page
- Lan, 2020, ALBERT: A lite BERT for self-supervised learning of language representations
- Li, 2022, Neural reality of argument structure constructions, 7410, 10.18653/v1/2022.acl-long.512 · DOI
- Linzen, 2016, Assessing the ability of LSTMs to learn syntax-sensitive dependencies, Transactions of the Association for Computational Linguistics, 4, 521, 10.1162/tacl_a_00115 · DOI · Paper page
- Liu, 2019, RoBERTa: A robustly optimized BERT pretraining approach, CoRR
- Lovering, 2021, Predicting inductive biases of pre-trained models
- Madabushi, 2020, CxGBERT: BERT meets construction grammar, 4020, 10.18653/v1/2020.coling-main.355 · DOI
- Mahowald, 2016, A meta-analysis of syntactic priming in language production, Journal of Memory and Language, 91, 5, 10.1016/j.jml.2016.03.009 · DOI · Paper page
- Marvin, 2018, Targeted syntactic evaluation of language models, 1192, 10.1016/j.jml.2016.03.009 · DOI · Paper page
- Merriam-Webster, 1989, Webster’s Dictionary of English Usage
- Nelson, 2004, The University of South Florida free association, rhyme, and word fragment norms, Behavior Research Methods, Instruments, & Computers, 36, 402, 10.3758/BF03195588 · DOI · Paper page
- Newman, 2021, Refining targeted syntactic evaluation of language models, 3710, 10.18653/v1/2021.naacl-main.290 · DOI
- Pham, 2021, Out of order: How important is the sequential order of words in a sentence in natural language understanding tasks?, 1145, 10.18653/v1/2021.findings-acl.98 · DOI · Paper page
- Pickering, 1998, The representation of verbs: Evidence from syntactic priming in language production, Journal of Memory and Language, 39, 633, 10.1006/jmla.1998.2592 · DOI · Paper page
- Pickering, 2013, Persistent structural priming and frequency effects during comprehension., Journal of Experimental Psychology: Learning, Memory, and Cognition, 39, 890, 10.1037/a0029181 · DOI
- Prasad, 2019, Using priming to uncover the organization of syntactic representations in neural language models, 66, 10.18653/v1/K19-1007 · DOI · Paper page
- Radford, 2019, Language models are unsupervised multitask learners, OpenAI blog, 1, 9
- Reitter, 2007, Against sequence priming: Evidence from constituents and distituents in corpus data
- Reitter, 2011, A computational cognitive model of syntactic priming, Cognitive Science, 35, 587, 10.1111/j.1551-6709.2010.01165.x · DOI · Paper page
- Reitter, 2014, Alignment and task success in spoken dialogue, Journal of Memory and Language, 76, 29, 10.1016/j.jml.2014.05.008 · DOI · Paper page
- Salazar, 2020, Masked language model scoring, 2699, 10.18653/v1/2020.acl-main.240 · DOI · Paper page
- Scheepers, 2003, Syntactic priming of relative clause attachments: Persistence of structural configuration in sentence production, Cognition, 89, 179, 10.1016/S0010-0277(03)00119-7 · DOI · Paper page
- Schrimpf, 2020, Artificial neural networks accurately predict language processing in the brain, bioRxiv
- Sinha, 2021, Masked language modeling and the distributional hypothesis: Order word matters pre-training for little, 2888, 10.18653/v1/2021.emnlp-main.230 · DOI · Paper page
- Sinha, 2021, UnNatural Language Inference, 7329, 10.18653/v1/2021.acl-long.569 · DOI · Paper page
- Tenney, 2019, What do you learn from context? Probing for sentence structure in contextualized word representations
- Tooley, 2014, On the parity of structural persistence in language production and comprehension, Cognition, 132, 101, 10.1016/j.cognition.2014.04.002 · DOI · Paper page
- Fox Tree, 1999, Building syntactic structure in speaking, Journal of Psycholinguistic Research, 28, 71, 10.1023/A:1023239604158 · DOI · Paper page
- Vamvas, 2021, On the limits of minimal pairs in contrastive evaluation, 58, 10.18653/v1/2021.blackboxnlp-1.5 · DOI
- van Schijndel, 2018, A neural model of adaptation in reading, pages 4704–pages 4710, 10.18653/v1/D18-1499 · DOI · Paper page
- Vaswani, 2017, Attention is all you need, 5998
- Warstadt, 2019, Investigating BERT’s knowledge of language: Five analysis methods with NPIs, 2870, 10.18653/v1/D19-1286 · DOI · Paper page
- Warstadt, 2020, BLiMP: A benchmark of linguistic minimal pairs for English, 351, 10.1162/tacl_a_00321 · DOI · Paper page
- Wheeldon, 2003, Phrase structure priming: A short-lived effect, Language and Cognitive Processes, 18, 431, 10.1080/01690960244000063 · DOI · Paper page
- White, 2021, A non-linear structural probe, 132
Cited by
Greshake, K., Abdelnabi, S., Mishra, S., Endres, C., Holz, T., & Fritz, M. (2023) · Proceedings of the 16th ACM Workshop on Artificial Intelligence and Security · doi:10.1145/3605764.3623985 · Cited by 667
“In this attack, the model spreads the injection to other models that may be reading those inbound messages.”
BackgroundLanguage Model Behavior: A Comprehensive Survey
Chang, T. A., & Bergen, B. K. (2024) · Computational Linguistics · doi:10.1162/coli_a_00492 · Cited by 105
“to repeat syntactic structures from the input context (Sinclair et al. 2022), with grammatical vs.”
BackgroundDo Large Language Models Know What Humans Know?
Trott, S., Jones, C., Chang, T., Michaelov, J., & Bergen, B. (2023) · Cognitive Science · doi:10.1111/cogs.13309 · Cited by 99
Hagendorff, T., Dasgupta, I., Binz, M., Chan, S. C. Y., Lampinen, A., Wang, J. X., Akata, Z., & Schulz, E. (2023) · arXiv · doi:10.48550/arxiv.2303.13988 · Cited by 68
Lampinen, A. (2024) · Computational Linguistics · doi:10.1162/coli_a_00525 · Cited by 32
“Similarly, LMs exhibit some analogous structural priming effects (Prasad, van Schijndel, and Linzen 2019; Sinclair et al. 2022), and might therefore likewise benefit from exposure to related structures before testing.”
Metadata
- Authors
- Sinclair, A., Jumelet, J., Zuidema, W., & Fernández, R.
- Venue
- Transactions of the Association for Computational Linguistics
- Publisher
- MIT Press
- Published
- 2022
- Volume
- 10
- Pages
- 1031-1050
- ISSN
- 2307-387X (ISSN-L)
- OpenAlex
- W4296564631
- Cited by
- 48
- References
- 18
- FWCI
- 3.47
- Type
- journal article
- Language
- en
- Open access
- diamond
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Structural Persistence in Language Models: Priming as a Window into Abstract Language Representations. https://app.cochat.ai/papers/10.1162/tacl_a_00504
- The paper
- Sinclair, A., Jumelet, J., Zuidema, W., & Fernández, R. (2022). Structural Persistence in Language Models: Priming as a Window into Abstract Language Representations. Transactions of the Association for Computational Linguistics. https://doi.org/10.1162/tacl_a_00504
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.