# Languages Still Left Behind: Toward a Better Multilingual Machine Translation Benchmark

## TL;DR

This paper critiques the FLORES+ multilingual MT benchmark, revealing quality issues, cultural biases, and evaluation vulnerabilities, and advocates for more representative, domain-neutral benchmarks to better assess real-world translation performance.

## Contribution

It uncovers critical shortcomings in FLORES+ and proposes guidelines for developing more effective, culturally neutral multilingual MT benchmarks.

## Key findings

- Many translations fall below 90% quality standard.
- Simple heuristics can inflate BLEU scores, exposing evaluation vulnerabilities.
- High-quality models perform poorly on FLORES+ but better on domain-neutral data.

## Abstract

Multilingual machine translation (MT) benchmarks play a central role in evaluating the capabilities of modern MT systems. Among them, the FLORES+ benchmark is widely used, offering English-to-many translation data for over 200 languages, curated with strict quality control protocols. However, we study data in four languages (Asante Twi, Japanese, Jinghpaw, and South Azerbaijani) and uncover critical shortcomings in the benchmark's suitability for truly multilingual evaluation. Human assessments reveal that many translations fall below the claimed 90% quality standard, and the annotators report that source sentences are often too domain-specific and culturally biased toward the English-speaking world. We further demonstrate that simple heuristics, such as copying named entities, can yield non-trivial BLEU scores, suggesting vulnerabilities in the evaluation protocol. Notably, we show that MT models trained on high-quality, naturalistic data perform poorly on FLORES+ while achieving significant gains on our domain-relevant evaluation set. Based on these findings, we advocate for multilingual MT benchmarks that use domain-general and culturally neutral source texts rely less on named entities, in order to better reflect real-world translation challenges.