Metraipolis › archive › dc_cc5d2709da51

Causal Interventions Reveal Typologically Organized Syntactic Mechanisms in Multilingual Language Models — arXiv:2608.28924

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.28924
License (as archived)
CC0-1.0
Archived
2026-10-17T08:06:48.532Z
SHA-256 of the archived text
7cf60120cb3a4d8ffa8b18d58c0f535682a6ce784adb55833628cc663840cae6
Stable URL of this page
https://metraipolis.com/documento/dc_cc5d2709da51

How to cite

Causal Interventions Reveal Typologically Organized Syntactic Mechanisms in Multilingual Language Models — arXiv:2608.28924 — arXiv (en), https://arxiv.org/abs/2608.28924 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cc5d2709da51 (Metraipolis archive dc_cc5d2709da51, sha256 7cf60120cb3a4d8ffa8b18d58c0f535682a6ce784adb55833628cc663840cae6)
BibTeX
@misc{dccc5d2709da51,
  title = {Causal Interventions Reveal Typologically Organized Syntactic Mechanisms in Multilingual Language Models — arXiv:2608.28924},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.28924},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cc5d2709da51; sha256 7cf60120cb3a4d8ffa8b18d58c0f535682a6ce784adb55833628cc663840cae6},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.28924. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Linguistic theory has long recognized cross-linguistic syntactic regularities, leading to claims that these similar structures are processed by similar mechanisms. However, this hypothesis has been difficult to test empirically due to our lack of fine-grained, manipulable access of human processing mechanisms. In this work, we take advantage of techniques from mechanistic interpretability to study such a question in multilingual LMs. We first isolate language-internal mechanisms before attempting to transfer them cross-lingually. Across four models and three well-studied constructions (subject--verb number agreement, anaphoric pronoun gender agreement, and filler--gap object extraction) we find consistent cross-lingual mechanism transfer. We further find transfer to be graded, with more transfer between more typologically similar languages. We believe our work provides novel hypotheses about cross-linguistic syntactic structures and multilingual processing, and more broadly shows how the study of language models can help inform linguistic theory.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.28924 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T07:47:52.662Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).