Beyond Bilingual Transfer: Multilingual Code-Switching in Instruction Tuning
Fuente:
arXiv
Guardado en:
| Autores principales: | , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866913169742495744 |
|---|---|
| author | Asano, Shunta Baek, Jeonghun Yamasaki, Toshihiko |
| author_facet | Asano, Shunta Baek, Jeonghun Yamasaki, Toshihiko |
| contents | Recent studies have shown that code-switching data (CSD), in which multiple languages are mixed within the same context, can improve cross-lingual transfer and multilingual alignment in large language models (LLMs). However, existing studies primarily focus on bilingual transfer between English and a target language, leaving multilingual settings involving three or more languages largely unexplored. In this work, we investigate multilingual code-switching instruction tuning across four languages: English, Japanese, Korean, and Chinese. We evaluate multilingual understanding on Belebele. Our experiments show that simple sentence-level multilingual CSD consistently improves average multilingual performance across all four languages, indicating that multilingual code-switching can be effective beyond bilingual transfer settings. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_29414 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Beyond Bilingual Transfer: Multilingual Code-Switching in Instruction Tuning Asano, Shunta Baek, Jeonghun Yamasaki, Toshihiko Computation and Language Artificial Intelligence Recent studies have shown that code-switching data (CSD), in which multiple languages are mixed within the same context, can improve cross-lingual transfer and multilingual alignment in large language models (LLMs). However, existing studies primarily focus on bilingual transfer between English and a target language, leaving multilingual settings involving three or more languages largely unexplored. In this work, we investigate multilingual code-switching instruction tuning across four languages: English, Japanese, Korean, and Chinese. We evaluate multilingual understanding on Belebele. Our experiments show that simple sentence-level multilingual CSD consistently improves average multilingual performance across all four languages, indicating that multilingual code-switching can be effective beyond bilingual transfer settings. |
| title | Beyond Bilingual Transfer: Multilingual Code-Switching in Instruction Tuning |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2605.29414 |