TransAlign: Machine Translation Encoders are Strong Word Aligners, Too

Benedikt Ebing, Christian Goldschmied, Goran Glavaš · 2025

Read TransAlign: Machine Translation Encoders are Strong Word Aligners, Too by Benedikt Ebing, Christian Goldschmied, Goran Glavaš (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-27337.txt.

Description

Benedikt Ebing, Christian Goldschmied, and Goran Glavaš turn a translation engine into a word-matching tool in this 2025 paper. Word alignment pairs each word in a sentence with its match in a translation. It is key when labels must move from one language to another for training. Most aligners come from plain language models, not translators.

Their tool, TransAlign, reads alignments from the encoder, the reading half, of a huge multilingual translator. Past work mined the translator's attention links and got weak results. TransAlign instead taps the encoder's inner states. In tests it beat popular aligners and other label-moving methods by a clear margin. That makes it a strong pick for carrying token labels across languages where training data is scarce.

Catalog record

CREATORBenedikt Ebing, Christian Goldschmied, Goran Glavaš
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-27337

Browse the full library · About the Tower of Babel