Metraipolis › archive › dc_601fccfb847b

What Travels Across Languages? Reading, Transfer and Erasure of Cross-Lingual Concept Subspaces in Bangla Idioms — arXiv:2611.11386

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.11386
License (as archived)
CC0-1.0
Archived
2026-11-10T18:44:38.015Z
SHA-256 of the archived text
97d70df44704fe9d8383f284c6dc7665372386541992e1f597ce952f7c0d81c1
Stable URL of this page
https://metraipolis.com/documento/dc_601fccfb847b

How to cite

What Travels Across Languages? Reading, Transfer and Erasure of Cross-Lingual Concept Subspaces in Bangla Idioms — arXiv:2611.11386 — arXiv (en), https://arxiv.org/abs/2611.11386 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_601fccfb847b (Metraipolis archive dc_601fccfb847b, sha256 97d70df44704fe9d8383f284c6dc7665372386541992e1f597ce952f7c0d81c1)
BibTeX
@misc{dc601fccfb847b,
  title = {What Travels Across Languages? Reading, Transfer and Erasure of Cross-Lingual Concept Subspaces in Bangla Idioms — arXiv:2611.11386},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.11386},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_601fccfb847b; sha256 97d70df44704fe9d8383f284c6dc7665372386541992e1f597ce952f7c0d81c1},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.11386. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Cross-lingual concept analysis requires distinguishing the information that a subspace retains from the classifiers and erasers that transfer through it. We study these operations in six frozen encoders using 10,359 Bangla (Bengali) idiom senses, eight input conditions, and figurative-class and sentiment labels. Class-mean directions provide the most consistent evidence of shared coarse meaning: in post hoc analyses, they outperform random-subspace, permuted-label and covariance-matched nulls in all 12 model-label tests under balanced accuracy. A prospective internal re-split supports idiom-alone sharing in 10 of 12 tests, spanning five encoders. The original confirmatory logistic-regression family rejects both planned nulls in 4 of 12 tests, where English directions gain only a few accuracy points above the majority-class floor (3.0-7.3). Classifier transfer depends on the input map and regularization: English-trained class-mean readouts beat both zero-shot nulls in 8 of 12 discovery tests, and stronger source-selected penalties improve transfer through logistic and LEACE directions. These readouts use Bangla-validation-selected layers and unlabeled target preprocessing. Input controls show greater coarse-class sharing for explicit glosses than for idiom counterparts. Reading, however, does not imply effective erasure: English-fitted LEACE removes less than one percentage point of Bangla figurative-class accuracy, compared with 4-11 points for Bangla-fitted LEACE; INLP shows a similar asymmetry. Pattern-filter analysis, covariance-weighted input maps and a binary Gaussian overlap bound explain how these operations can diverge. The analogous multiclass ordering is an empirical diagnostic, observed in 11 of 12 cells. The results support an estimator-sensitive account of cross-lingual concept sharing and motivate evaluating reading, classifier transfer, erasure and overlap separately.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG); Machine Learning (stat.ML).
Pagina del lavoro: https://arxiv.org/abs/2611.11386 — 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 2026-11-10T18:37:52.438Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).