pith. sign in

arxiv: 2504.15521 · v1 · pith:S4UB4TFUnew · submitted 2025-04-22 · 💻 cs.CL

The Bitter Lesson Learned from 2,000+ Multilingual Benchmarks

classification 💻 cs.CL
keywords benchmarksmultilingualhumanbenchmarkingcorrelationscountriesenglishevaluation
0
0 comments X
read the original abstract

As large language models (LLMs) continue to advance in linguistic capabilities, robust multilingual evaluation has become essential for promoting equitable technological progress. This position paper examines over 2,000 multilingual (non-English) benchmarks from 148 countries, published between 2021 and 2024, to evaluate past, present, and future practices in multilingual benchmarking. Our findings reveal that, despite significant investments amounting to tens of millions of dollars, English remains significantly overrepresented in these benchmarks. Additionally, most benchmarks rely on original language content rather than translations, with the majority sourced from high-resource countries such as China, India, Germany, the UK, and the USA. Furthermore, a comparison of benchmark performance with human judgments highlights notable disparities. STEM-related tasks exhibit strong correlations with human evaluations (0.70 to 0.85), while traditional NLP tasks like question answering (e.g., XQuAD) show much weaker correlations (0.11 to 0.30). Moreover, translating English benchmarks into other languages proves insufficient, as localized benchmarks demonstrate significantly higher alignment with local human judgments (0.68) than their translated counterparts (0.47). This underscores the importance of creating culturally and linguistically tailored benchmarks rather than relying solely on translations. Through this comprehensive analysis, we highlight six key limitations in current multilingual evaluation practices, propose the guiding principles accordingly for effective multilingual benchmarking, and outline five critical research directions to drive progress in the field. Finally, we call for a global collaborative effort to develop human-aligned benchmarks that prioritize real-world applications.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 7 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Physics-R1: An Audited Olympiad Corpus and Recipe for Visual Physics Reasoning

    cs.CL 2026-05 unverdicted novelty 7.0

    Audited olympiad corpus and Physics-R1 recipe improve 8B VLM by up to 18 points on held-out physics problems while exposing contamination in prior evals.

  2. The Translation Tax Is Not a Scalar: A Counterfactual Audit of English-Source Cue Inheritance in Chinese Multilingual Benchmarks

    cs.CL 2026-05 conditional novelty 7.0

    An empirical audit shows the translation tax in English-to-Chinese benchmarks is estimator- and item-dependent rather than a single scalar effect, with a residue dose-response in naturalization tests.

  3. OmniMol: Transferring Particle Physics Knowledge to Molecular Dynamics with Point-Edge Transformers

    physics.chem-ph 2026-01 unverdicted novelty 7.0

    OmniMol transfers a billion-jet pre-trained PET foundation model from HEP to molecular dynamics via an interaction-matrix attention bias, delivering strong performance on the oMol dataset with minimal fine-tuning and ...

  4. XLGoBench: Detecting cross-lingual skill gaps with algorithmic tasks

    cs.CL 2026-05 unverdicted novelty 6.0

    XLGoBench applies algorithmic tasks to expose persistent cross-lingual performance gaps in state-of-the-art LLMs.

  5. Instructions Shape Production of Language, not Processing

    cs.CL 2026-05 unverdicted novelty 6.0

    Instructions trigger a production-centered mechanism in language models, with task-specific information stable in input tokens but varying strongly in output tokens and correlating with behavior.

  6. CulturALL: Benchmarking Multilingual and Multicultural Competence of LLMs on Grounded Tasks

    cs.CL 2026-04 unverdicted novelty 6.0

    CulturALL is a new benchmark where even the best LLMs reach only 44.48% accuracy on grounded multilingual and multicultural tasks.

  7. Instructions Shape Production of Language, not Processing

    cs.CL 2026-05 unverdicted novelty 5.0

    Instructions primarily shape the production stage of language models rather than the processing stage, with task-specific information and causal effects stronger in output tokens than input tokens.