# Controlling Language Confusion in Multilingual LLMs

arXiv: https://arxiv.org/abs/2505.19116 · HTML: https://mlsanity.com/paper/2505.19116/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 14 of its 28 references that are in the map.
- First posted on arXiv 25 May 2025.

## Builds on (most cited first, 14 of 14)

- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [ORPO: Monolithic Preference Optimization without Reference Model](https://mlsanity.com/paper/2403.07691/) (2024)
- [How Close is ChatGPT to Human Experts? Comparison Corpus, Evaluation, and Detection](https://mlsanity.com/paper/2301.07597/) (2023)
- [Massively Multilingual Neural Machine Translation in the Wild: Findings and Challenges](https://mlsanity.com/paper/1907.05019/) (2019)
- [Tower: An Open Multilingual Large Language Model for Translation-Related Tasks](https://mlsanity.com/paper/2402.17733/) (2024)
- [SeaEval for Multilingual Foundation Models: From Cross-Lingual Alignment to Cultural Reasoning](https://mlsanity.com/paper/2309.04766/) (2023)
- [Understanding and Mitigating Language Confusion in LLMs](https://mlsanity.com/paper/2406.20052/) (2024)
- [Command A: An Enterprise-Ready Large Language Model](https://mlsanity.com/paper/2504.00698/) (2025)
- [HAE-RAE Bench: Evaluation of Korean Knowledge in Language Models](https://mlsanity.com/paper/2309.02706/) (2023)
- [Linguistic Generalizability of Test-Time Scaling in Mathematical Reasoning](https://mlsanity.com/paper/2502.17407/) (2025)
- [MM-Eval: A Multilingual Meta-Evaluation Benchmark for LLM-as-a-Judge and Reward Models](https://mlsanity.com/paper/2410.17578/) (2024)
- [Code-Switching Curriculum Learning for Multilingual Transfer in LLMs](https://mlsanity.com/paper/2411.02460/) (2024)
- [Investigating and Scaling up Code-Switching for Multilingual Language Model Pre-Training](https://mlsanity.com/paper/2504.01801/) (2025)
- [Large Language Models are Easily Confused: A Quantitative Metric, Security Implications and Typological Analysis](https://mlsanity.com/paper/2410.13237/) (2024)

## Built on by (most cited first, 3 of 3)

- [When Meanings Meet: Investigating the Emergence and Quality of Shared Concept Spaces during Multilingual Language Model Training](https://mlsanity.com/paper/2601.22851/) (2026)
- [Language Confusion Gate: Language-Aware Decoding Through Model Self-Distillation](https://mlsanity.com/paper/2510.17555/) (2025)
- [ChiEngMixBench: Evaluating Large Language Models on Expert-Style Chinese-English Terminology Mixing](https://mlsanity.com/paper/2601.16217/) (2026)

## Abstract (arXiv)

Large language models often suffer from language confusion, a phenomenon in which responses are partially or entirely generated in unintended languages. This critically degrades the user experience, especially in low-resource settings. We hypothesize that this issue stems from limitations in conventional fine-tuning objectives, such as supervised learning, which optimize the likelihood of correct tokens without explicitly penalizing undesired outputs such as cross-lingual mixing. Analysis of loss trajectories during pretraining further reveals that models fail to distinguish between monolingual and language-mixed texts, highlighting the absence of inherent pressure to avoid such confusion. In this work, we apply ORPO, which adds penalties for unwanted output styles to standard SFT, effectively suppressing language-confused generations. ORPO maintains strong language consistency, even under high decoding temperatures, while preserving general QA performance. Our findings suggest that incorporating appropriate penalty terms can effectively mitigate language confusion in multilingual models, particularly in low-resource scenarios.