Causal Interventions Reveal Shared Structure Across English Filler-Gap Constructions

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Boguraev, Sasha, Potts, Christopher, Mahowald, Kyle
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915523440148480
author Boguraev, Sasha
Potts, Christopher
Mahowald, Kyle
author_facet Boguraev, Sasha
Potts, Christopher
Mahowald, Kyle
contents Language Models (LMs) have emerged as powerful sources of evidence for linguists seeking to develop theories of syntax. In this paper, we argue that causal interpretability methods, applied to LMs, can greatly enhance the value of such evidence by helping us characterize the abstract mechanisms that LMs learn to use. Our empirical focus is a set of English filler-gap dependency constructions (e.g., questions, relative clauses). Linguistic theories largely agree that these constructions share many properties. Using experiments based in Distributed Interchange Interventions, we show that LMs converge on similar abstract analyses of these constructions. These analyses also reveal previously overlooked factors -- relating to frequency, filler type, and surrounding context -- that could motivate changes to standard linguistic theory. Overall, these results suggest that mechanistic, internal analyses of LMs can push linguistic theory forward.
format Preprint
id arxiv_https___arxiv_org_abs_2505_16002
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Causal Interventions Reveal Shared Structure Across English Filler-Gap Constructions
Boguraev, Sasha
Potts, Christopher
Mahowald, Kyle
Computation and Language
Artificial Intelligence
Language Models (LMs) have emerged as powerful sources of evidence for linguists seeking to develop theories of syntax. In this paper, we argue that causal interpretability methods, applied to LMs, can greatly enhance the value of such evidence by helping us characterize the abstract mechanisms that LMs learn to use. Our empirical focus is a set of English filler-gap dependency constructions (e.g., questions, relative clauses). Linguistic theories largely agree that these constructions share many properties. Using experiments based in Distributed Interchange Interventions, we show that LMs converge on similar abstract analyses of these constructions. These analyses also reveal previously overlooked factors -- relating to frequency, filler type, and surrounding context -- that could motivate changes to standard linguistic theory. Overall, these results suggest that mechanistic, internal analyses of LMs can push linguistic theory forward.
title Causal Interventions Reveal Shared Structure Across English Filler-Gap Constructions
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2505.16002