Assessing the Capacity of Transformer to Abstract Syntactic Representations: A Contrastive Analysis Based on Long-distance Agreement
This work addresses the problem of understanding syntactic generalization in neural language models for researchers in computational linguistics, though it is incremental in nature.
The study investigated how transformers handle two distinct long-distance agreement phenomena in French, revealing that existing agreement tasks contain confounders and that transformers process these agreements in ways consistent with theoretical linguistics.
The long-distance agreement, evidence for syntactic structure, is increasingly used to assess the syntactic generalization of Neural Language Models. Much work has shown that transformers are capable of high accuracy in varied agreement tasks, but the mechanisms by which the models accomplish this behavior are still not well understood. To better understand transformers' internal working, this work contrasts how they handle two superficially similar but theoretically distinct agreement phenomena: subject-verb and object-past participle agreement in French. Using probing and counterfactual analysis methods, our experiments show that i) the agreement task suffers from several confounders which partially question the conclusions drawn so far and ii) transformers handle subject-verb and object-past participle agreements in a way that is consistent with their modeling in theoretical linguistics.