Beyond Monolingual Assumptions: A Survey of Code-Switched NLP in the Era of Large Language Models

2025-10-08

AI Summary

Code-switching (CSW)—the intra-sentential mixing of languages/scripts—remains a critical bottleneck for deploying large language models (LLMs) in multilingual societies, manifesting as poor comprehension of mixed inputs, scarce training data, and biased evaluation. This paper presents the first systematic survey of CSW-NLP research in the LLM era, covering five research directions, twelve tasks, and over thirty datasets. We propose a tripartite development framework: inclusive data construction, fair evaluation protocols, and linguistics-informed modeling. Through architectural analysis, taxonomy of training strategies, and integration of evaluation methodologies, we build the first open-source CSW resource repository supporting 80+ languages. Empirical analysis reveals significant limitations of mainstream LLMs in cross-lingual semantic alignment and syntactic consistency. Our work provides both theoretical foundations and practical benchmarks for trustworthy, culturally aware multilingual AI.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Language and Vision

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Code-switching (CSW), the alternation of languages and scripts within a single utterance, remains a fundamental challenge for multiling ual NLP, even amidst the rapid advances of large language models (LLMs). Most LLMs still struggle with mixed-language inputs, limited CSW datasets, and evaluation biases, hindering deployment in multilingual societies. This survey provides the first comprehensive analysis of CSW-aware LLM research, reviewing otal{unique_references} studies spanning five research areas, 12 NLP tasks, 30+ datasets, and 80+ languages. We classify recent advances by architecture, training strategy, and evaluation methodology, outlining how LLMs have reshaped CSW modeling and what challenges persist. The paper concludes with a roadmap emphasizing the need for inclusive datasets, fair evaluation, and linguistically grounded models to achieve truly multilingual intelligence. A curated collection of all resources is maintained at https://github.com/lingo-iitgn/awesome-code-mixing/.

Problem

• Addressing code-switching challenges in multilingual NLP with large language models
• Overcoming limited datasets and evaluation biases in mixed-language processing
• Developing linguistically grounded models for true multilingual intelligence

Innovation

• Surveyed code-switched NLP with large language models
• Classified advances by architecture and training strategy
• Proposed roadmap for inclusive datasets and evaluation