Code-mixing and Code switching in low-resourced languages

Research on code-mixing and code-switching is progressing with the introduction of dedicated benchmarks for tasks like sentiment analysis, question answering, machine translation, and named entity recognition for mixed-language text. The robust cross-lingual representation and knowledge generalizability of transformer-based multilingual large language models, like mBERT, XLM-R, OpenAI GPT models, and classical machine learning methods like support vector machine, Random Forest etc. are currently being exploited to address the challenge of code-mixing and code-switching. The current efforts are significant in building and advancing emerging LLMs and generative AI systems addressing real-world language modelling. In this comparison, we capture some research efforts that have contributed models, methods, and evaluation benchmarks to support the domain.