dorsal/arxiv
View SchemaMultilinguality as Sense Adaptation
| Authors | Jan Christian Blaise Cruz, David Ifeoluwa Adelani, Alham Fikri Aji |
|---|---|
| Categories | |
| ArXiv ID | 2601.10310vv1 |
| URL | https://arxiv.org/abs/2601.10310 |
| License | http://creativecommons.org/licenses/by-sa/4.0/ |
Abstract
We approach multilinguality as sense adaptation: aligning latent meaning representations across languages rather than relying solely on shared parameters and scale. In this paper, we introduce SENse-based Symmetric Interlingual Alignment (SENSIA), which adapts a Backpack language model from one language to another by explicitly aligning sense-level mixtures and contextual representations on parallel data, while jointly training a target-language language modeling loss to preserve fluency. Across benchmarks on four typologically diverse languages, SENSIA generally outperforms comparable multilingual alignment methods and achieves competitive accuracy against monolingual from-scratch baselines while using 2-4x less target-language data. Analyses of learned sense geometry indicate that local sense topology and global structure relative to English are largely preserved, and ablations show that the method is robust in terms of design and scale.
{
"annotation_id": "e060726e-7173-4feb-964e-d373851db9d7",
"date_created": "2026-02-17T05:53:23.353000Z",
"date_modified": "2026-02-17T05:53:23.353000Z",
"file_hash": "731a31b1c2f469281fd3503910232e5228d0f82f43193296a26a98bd88191af3",
"private": false,
"record": {
"abstract": "We approach multilinguality as sense adaptation: aligning latent meaning representations across languages rather than relying solely on shared parameters and scale. In this paper, we introduce SENse-based Symmetric Interlingual Alignment (SENSIA), which adapts a Backpack language model from one language to another by explicitly aligning sense-level mixtures and contextual representations on parallel data, while jointly training a target-language language modeling loss to preserve fluency. Across benchmarks on four typologically diverse languages, SENSIA generally outperforms comparable multilingual alignment methods and achieves competitive accuracy against monolingual from-scratch baselines while using 2-4x less target-language data. Analyses of learned sense geometry indicate that local sense topology and global structure relative to English are largely preserved, and ablations show that the method is robust in terms of design and scale.",
"arxiv_id": "2601.10310",
"authors": [
"Jan Christian Blaise Cruz",
"David Ifeoluwa Adelani",
"Alham Fikri Aji"
],
"categories": [
"cs.CL"
],
"license": "http://creativecommons.org/licenses/by-sa/4.0/",
"title": "Multilinguality as Sense Adaptation",
"url": "https://arxiv.org/abs/2601.10310",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "c95ac034-9b2b-496f-8ab0-574477d5677a",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}