Causal Analysis of Syntactic Agreement Neurons in Multilingual Language Models

Aaron Mueller, Yu Xia, Tal Linzen

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    Structural probing work has found evidence for latent syntactic information in pre-trained language models. However, much of this analysis has focused on monolingual models, and analyses of multilingual models have employed correlational methods that are confounded by the choice of probing tasks. In this study, we causally probe multilingual language models (XGLM and multilingual BERT) as well as monolingual BERT-based models across various languages; we do this by performing counterfactual perturbations on neuron activations and observing the effect on models' subjectverb agreement probabilities. We observe where in the model and to what extent syntactic agreement is encoded in each language. We find significant neuron overlap across languages in autoregressive multilingual language models, but not masked language models. We also find two distinct layer-wise effect patterns and two distinct sets of neurons used for syntactic agreement, depending on whether the subject and verb are separated by other tokens. Finally, we find that behavioral analyses of language models are likely underestimating how sensitive masked language models are to syntactic information.

    Original languageEnglish (US)
    Title of host publicationCoNLL 2022 - 26th Conference on Computational Natural Language Learning, Proceedings of the Conference
    PublisherAssociation for Computational Linguistics (ACL)
    Pages95-109
    Number of pages15
    ISBN (Electronic)9781959429074
    StatePublished - 2022
    Event26th Conference on Computational Natural Language Learning, CoNLL 2022 collocated and co-organized with EMNLP 2022 - Abu Dhabi, United Arab Emirates
    Duration: Dec 7 2022Dec 8 2022

    Publication series

    NameCoNLL 2022 - 26th Conference on Computational Natural Language Learning, Proceedings of the Conference

    Conference

    Conference26th Conference on Computational Natural Language Learning, CoNLL 2022 collocated and co-organized with EMNLP 2022
    Country/TerritoryUnited Arab Emirates
    CityAbu Dhabi
    Period12/7/2212/8/22

    ASJC Scopus subject areas

    • Artificial Intelligence
    • Human-Computer Interaction
    • Linguistics and Language

    Fingerprint

    Dive into the research topics of 'Causal Analysis of Syntactic Agreement Neurons in Multilingual Language Models'. Together they form a unique fingerprint.

    Cite this