DATASHI: A Parallel English–Tashlhiyt Corpus for Orthography Normalization and Low-Resource Language Processing.

2026

Read DATASHI: A Parallel English–Tashlhiyt Corpus for Orthography Normalization and Low-Resource Language Processing. (2026) 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-2603-21571.txt.

Description

Monir and Baou introduce DATASHI, a parallel English-Tashlhiyt corpus for a low-resource Amazigh language. The corpus holds 5,000 sentence pairs. A 1,500-sentence subset pairs expert-standardized text with raw user-written versions, which supports study of spelling variation and normalization. They benchmark several large language models on the corpus in zero-shot and few-shot modes. Gemini-2.5-Pro posted the lowest word and character error rates. A fine-grained error analysis shows which Tashlhiyt sound classes, like geminates and pharyngeals, trip up each model.

Catalog record

YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34925
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.21571

Browse the full library · About the Tower of Babel