# Extending Item Response Theory for Efficient and Meaningful Multilingual Evaluation

arXiv: https://arxiv.org/abs/2606.15643 · HTML: https://mlsanity.com/paper/2606.15643/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 19 of its 32 references that are in the map.
- First posted on arXiv 14 Jun 2026.

## Builds on (most cited first, 19 of 19)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Replacing Judges with Juries: Evaluating LLM Generations with a Panel of Diverse Models](https://mlsanity.com/paper/2404.18796/) (2024)
- [The Belebele Benchmark: a Parallel Reading Comprehension Dataset in 122 Language Variants](https://mlsanity.com/paper/2308.16884/) (2023)
- [tinyBenchmarks: evaluating LLMs with fewer examples](https://mlsanity.com/paper/2402.14992/) (2024)
- [Aya Model: An Instruction Finetuned Open-Access Multilingual Language Model](https://mlsanity.com/paper/2402.07827/) (2024)
- [MEGA: Multilingual Evaluation of Generative AI](https://mlsanity.com/paper/2303.12528/) (2023)
- [Composable Effects for Flexible and Accelerated Probabilistic Programming in NumPyro](https://mlsanity.com/paper/1912.11554/) (2019)
- [A Survey on Multilingual Large Language Models: Corpora, Alignment, and Bias](https://mlsanity.com/paper/2404.00929/) (2024)
- [Lost in Benchmarks? Rethinking Large Language Model Benchmarking with Item Response Theory](https://mlsanity.com/paper/2505.15055/) (2025)
- [Leveraging LLM-Respondents for Item Evaluation: a Psychometric Analysis](https://mlsanity.com/paper/2407.10899/) (2024)
- [The Bitter Lesson Learned from 2,000+ Multilingual Benchmarks](https://mlsanity.com/paper/2504.15521/) (2025)
- [P-MMEval: A Parallel Multilingual Multitask Benchmark for Consistent Evaluation of LLMs](https://mlsanity.com/paper/2411.09116/) (2024)
- [Adaptive Testing for LLM Evaluation: A Psychometric Alternative to Static Benchmarks](https://mlsanity.com/paper/2511.04689/) (2025)
- [Item Response Theory -- A Statistical Framework for Educational and Psychological Measurement](https://mlsanity.com/paper/2108.08604/) (2021)
- [Do LLMs Give Psychometrically Plausible Responses in Educational Assessments?](https://mlsanity.com/paper/2506.09796/) (2025)
- [Mind the Gap... or Not? How Translation Errors and Evaluation Details Skew Multilingual Results](https://mlsanity.com/paper/2511.05162/) (2025)
- [Translation Errors Significantly Impact Low-Resource Languages in Cross-Lingual Learning](https://mlsanity.com/paper/2402.02080/) (2024)
- [Estimation Methods for Item Factor Analysis: An Overview](https://mlsanity.com/paper/2004.07579/) (2020)

## Built on by (most cited first, 2 of 2)

- [Every Wrong Answer Counts: Option-Level Psychometrics for LLM Multiple-Choice Benchmarks](https://mlsanity.com/paper/2608.02966/) (2026)
- [Are We Testing Too Much? A Survey of Benchmark Compression for Efficient Evaluation of Large Language Models](https://mlsanity.com/paper/2610.18248/) (2026)

## Abstract (arXiv)

Multilingual benchmarks are central to evaluating large language models (LLMs) across languages, but they suffer from three issues: exhaustive evaluation scales linearly with the number of languages, automatic translation introduces errors that are easily missed at scale, and some items conflate general and culture-specific knowledge. We address all three with a unified statistical framework, Multilingual-IRT, which extends Item Response Theory with per-language difficulty deviations, split discriminability separating content from language effects, and per-language ability residuals. Fitting Multilingual-IRT on 25 LLMs across 29 languages of MMLU-Pro-X, we show that its fitted parameters support three practical applications: predicting unobserved (item, LLM, language) instances with 11-16% lower binary cross-entropy than the strongest accuracy-based baseline, surfacing candidate translation errors distributed across all 28 non-English languages, whereas accuracy-based baselines concentrate detections in a few languages, and recovering culture-specific items that accuracy-based baselines miss.