Beyond Literal Token Overlap: Token Alignability for Multilinguality

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F25%3A10511589" target="_blank" >RIV/00216208:11320/25:10511589 - isvavai.cz</a>

Result on the web

<a href="https://aclanthology.org/2025.naacl-short.63" target="_blank" >https://aclanthology.org/2025.naacl-short.63</a>

DOI - Digital Object Identifier

—

Result language

angličtina

Original language name

Beyond Literal Token Overlap: Token Alignability for Multilinguality

Original language description

Previous work has considered token overlap, or even similarity of token distributions, as predictors for multilinguality and cross-lingual knowledge transfer in language models. However, these very literal metrics assign large distances to language pairs with different scripts, which can nevertheless show good cross-linguality. This limits the explanatory strength of token overlap for knowledge transfer between language pairs that use distinct scripts or follow different orthographic conventions. In this paper, we propose subword token alignability as a new way to understand the impact and quality of multilingual tokenisation. In particular, this metric predicts multilinguality much better when scripts are disparate and the overlap of literal tokens is low. We analyse this metric in the context of both encoder and decoder models, look at data size as a potential distractor, and discuss how this insight may be applied to multilingual tokenisation in future work. We recommend our subword token alignabil

Czech name

—

Czech description

—

Type

D - Article in proceedings

CEP classification

—

OECD FORD branch

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Project

—

Continuities

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Publication year

2025

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection

Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 2: Short Papers)

ISBN

979-8-89176-190-2

ISSN

—

e-ISSN

—

Number of pages

12

Pages from-to

756-767

Publisher name

Association for Computational Linguistics

Place of publication

Kerrville, TX, USA

Event location

Albuquerque, NM, USA

Event date

Apr 29, 2025

Type of event by nationality

WRD - Celosvětová akce

UT code for WoS article

—