Causal Analysis of Syntactic Agreement Neurons in Multilingual Language Models

Proceedings of the 26th Conference on Computational Natural Language Learning (CoNLL), pages 95 - 109 December 7-8, 2022 ©2022 Association for Computational Linguistics Causal Analysis of Syntactic Agreement Neurons in Multilingual Language Models Aaron MuellerS, Yu XiaC, Tal LinzenC SJohns Hopkins University CNew York University amueller@jhu.edu, yx1675@nyu.edu, linzen@nyu.edu Abstract Structural probing work has found evidence for latent syntactic information in pre-trained language models. However, much of this anal- ysis 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 vari- ous languages; we do this by performing coun- terfactual perturbations on neuron activations and observing the effect on models’ subject- verb agreement probabilities. We observe where in the model and to what extent syn- tactic agreement is encoded in each language. We find significant neuron overlap across lan- guages 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 syn- tactic agreement, depending on whether the subject and verb are separated by other tokens. Finally, we find that behavioral analyses of lan- guage models are likely underestimating how sensitive masked language models are to syn- tactic information. 1 Introduction Syntactic information is necessary for robust gener- alization in natural language processing tasks (for a case study using the natural language inference task, see McCoy et al. 2019). The success of pre- trained language models (LMs) such as RoBERTa (Liu et al., 2019) and GPT-3 (Brown et al., 2020) in many NLP tasks has prompted hypotheses that they accomplish their performance through struc- tural representations induced during pre-training, rather than only lexical or positional represen- tations (Manning et al., 2020); behavioral evi- dence for LMs’ syntactic abilities has been found in masked LMs (MLMs; Warstadt et al., 2020; Warstadt and Bowman, 2020; Goldberg, 2019) and autoregressive LMs (ALMs; Hu et al., 2020). Ev- idence for structural representations has been re- ported for multilingual pre-trained LMs (Goldberg, 2019; Mueller et al., 2020) and in sequence-to- sequence models (Mueller et al., 2022). Despite efforts to understand the structural infor- mation encoded by pre-trained LMs (Hewitt and Manning, 2019; Chi et al., 2020; Elazar et al., 2021; Ravfogel et al., 2021; Finlayson et al., 2021; inter alia), it remains unclear how and where multilin- gual models encode this information. Most multi- lingual probing studies are correlational and use dependency parsing or labeling as a proxy task in- dicative of syntactic information (Chi et al., 2020; Sta´nczak et al., 2022). This is problematic: Models do not need structural or word order information to achieve high performance on dependency labeling (Sinha et al., 2021), and training a parametric prob- ing classifier introduces many confounds (Hewitt and Liang, 2019; Antverg and Belinkov, 2022). Causal probing, however, enables non- parametric analyses of models through coun- terfactual interventions on inputs …

Content truncated to 500 words