Published March 5, 2026 | Version 2.0.0
Dataset Open

The English Trap: Gender Bias and Grammatical Information Loss Through English-Influenced Universal Representations in Multilingual NMT

  • 1. ROR icon University of Western Macedonia

Description

Dataset and code accompanying the paper "The English Trap" (2026). Contains 112 annotated sentences across 3 MT systems and 2 translation directions (Spanish→Greek, Greek→Spanish), with gender bias annotations, justifications, error type classifications, full pivot language translations, back-translations, and exact-match scores for 16 candidate pivot languages. Version 2.0: complete dataset replacing v1 which had incomplete pivot scores.

Files

EL_ES_deepl_classic.csv

Files (791.5 kB)

Name Size Download all
md5:b5d31a7b72594107422a7a70e6063879
194.0 kB Download
md5:24af7534a959aedf9443bf844343a163
84.7 kB Preview Download
md5:0bd2a94df1bdc5b7ba5dabe0a2cf575a
83.3 kB Preview Download
md5:264270e8ec07681e0900545bc67d983a
84.8 kB Preview Download
md5:ccec3339ba1949c73ef925d0ea10d483
109.9 kB Preview Download
md5:fbe4f6dc82106d4064d20ca189c0696b
113.0 kB Preview Download
md5:9019a1588438cc3b6211df80a6e00998
109.8 kB Preview Download
md5:71223eab237f2a336a3c8af62e69490f
11.8 kB Download

Additional details

Related works