Investigating and Scaling up Code-Switching for Multilingual Language Model Pre-Training (2025.findings-acl)
Copied to clipboard
Zhijun Wang, Jiahuan Li, Hao Zhou, Rongxiang Weng, Jingang Wang, Xin Huang, Xue Han, Junlan Feng, Chao Deng, Shujian Huang
| Challenge: | Large language models (LLMs) exhibit remarkable multilingual capabilities despite the extreme language imbalance in the pre-training data. |
| Approach: | They investigate the existence of code-switching in the pre-training corpus and categorize it into four types within two quadrants. |
| Outcome: | The proposed approach improves performance across benchmarks and representation space. |
Similar Papers
Code-Mixed Probes Show How Pre-Trained Models Generalise on Code-Switched Text (2024.lrec-main)
Copied to clipboard
| Challenge: | Code-switching is a prevalent linguistic phenomenon in which multilingual individuals seamlessly alternate between languages. |
| Approach: | They propose to use pre-trained language models to generalise to code-switched text . they use a dataset of well-formed naturalistic code-witched texts and parallel translations into the source languages to examine their results. |
| Outcome: | The proposed model generalises to code-switched text, shedding light on their ability to generalise representations to CS corpora. |
The Role of Mixed-Language Documents for Multilingual Large Language Model Pretraining (2026.acl-long)
Copied to clipboard
Jiandong Shao, Raphael Tang, Crystina Zhang, Karin Sevegnani, Pontus Stenetorp, Jianfei Yang, Yao Lu
| Challenge: | Existing research suggests that multilingual large language models can achieve impressive cross-lingual understanding despite largely monolingual pretraining. |
| Approach: | They compare a monolingual-only corpus with a standard web corpus that removes all multilingual documents and then retrain the models from scratch under controlled conditions. |
| Outcome: | The results show that removing bilingual data causes translation performance to drop 56% in BLEU, whereas code-switching contributes minimally. |
Evaluating Code-Switching Translation with Large Language Models (2024.lrec-main)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have shown they can match or surpass finetuned models on many natural language processing tasks. |
| Approach: | They propose to use in-context learning and pivot translation to improve code-switching translation. |
| Outcome: | The proposed models show strong ability for cross-lingual understanding in a code-switching setting. |
Exploring Enhanced Code-Switched Noising for Pretraining in Neural Machine Translation (2023.findings-eacl)
Copied to clipboard
| Challenge: | Multilingual pretraining approaches to denoise synthetic code-switched data have shown that they generate the noise using non-contextual, one-to-one word translations obtained from lexicons. |
| Approach: | They propose an approach where contextual, many-to-many word translations are generated using a ‘base’ NMT model. |
| Outcome: | The proposed approach improves on 3 different language families and shows that small models can perform better than massive models like mBART50 and mRASP2 . |
Beyond Monolingual Assumptions: A Survey on Code-Switched NLP in the Era of Large Language Models across Modalities (2026.acl-long)
Copied to clipboard
| Challenge: | Amidst the rapid advances of large language models, most LLMs struggle with mixed-language inputs, limited Code-switching datasets, and evaluation biases. |
| Approach: | They propose a roadmap for inclusive datasets, fair evaluation, and linguistically grounded models to achieve truly multilingual intelligence. |
| Outcome: | The proposed frameworks are based on 327 studies spanning five research areas, 15+ NLP tasks, 30+ datasets, and 80+ languages. |
Can Code-Switched Texts Activate a Knowledge Switch in LLMs? A Case Study on English-Korean Code-Switching (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Recent large language models (LLMs) demonstrate multilingual abilities, yet they are English-centric due to dominance of English in training corpora. |
| Approach: | They propose to use a synthetic English-korean CS question-answering dataset to investigate this potential. |
| Outcome: | The proposed model can activate, identify and leverage knowledge for reasoning in low-resource languages. |
Scaling Laws for Code: Every Programming Language Matters (2026.findings-acl)
Copied to clipboard
Jian Yang, Shuyue Guo, Linzheng Chai, Wei Zhang, Aishan Liu, Chuan Hao, Zhoujun Li, Xin Zhao, Xianglong Liu, Weifeng Lv, Bryan Dai
| Challenge: | Existing studies focus on language-agnostic settings, neglecting the inherently multilingual nature of modern software development. |
| Approach: | They propose a proportion-dependent scaling law that prioritizes high-utility languages . they propose PLs to have varying effects during pre-training that affect model performance . |
| Outcome: | The proposed scaling law is based on 1000+ experiments across multiple languages and models. |
Lost in the Mix: Evaluating LLM Understanding of Code-Switched Text (2026.acl-long)
Copied to clipboard
| Challenge: | Code-switching (CSW) is widespread in multilingual communities and increasingly prevalent in online content. |
| Approach: | They propose a pipeline for producing linguistically grounded CSW variants of established benchmarks across five typologically diverse languages. |
| Outcome: | The proposed model sets show that inserting non-English tokens into English reduces accuracy on comprehension and reasoning benchmarks, whereas embedding English into non- English contexts often improves it. |
Code-Switching Curriculum Learning for Multilingual Transfer in LLMs (2025.findings-acl)
Copied to clipboard
| Challenge: | Large language models (LLMs) exhibit near human-level performance in various tasks, but performance drops after a handful of high-resource languages due to the imbalance in pre-training data. |
| Approach: | They propose a code-switching curriculum learning model to enhance cross-lingual transfer for LLMs by progressively training models with a curriculum consisting of token-level code-changing, sentence-level codeswitching, and monolingual corpora. |
| Outcome: | The proposed model improves language transfer to Korean, with significant gains in Japanese and Indonesian . the proposed model mitigates spurious correlations between language resources and safety alignment . |
Multilingual Large Language Models Are Not (Yet) Code-Switchers (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing multilingual Large Language Models are not specifically trained with objectives for managing code-switching scenarios. |
| Approach: | They propose to use multilingual Large Language Models to perform sentiment analysis, machine translation, summarization and word-level language identification to compare their performance to fine-tuned models of much smaller scales. |
| Outcome: | The proposed models show that they underperform in comparison to fine-tuned models of much smaller scales. |