@inproceedings{trinley-etal-2025-language,
    title = "What Language(s) Does Aya-23 Think In? How Multilinguality Affects Internal Language Representations",
    author = "Trinley, Katharina A. T. T.  and
      Nakai, Toshiki  and
      Anikina, Tatiana  and
      Baeumel, Tanja",
    editor = "Das, Sudhansu Bala  and
      Mishra, Pruthwik  and
      Singh, Alok  and
      Muhammad, Shamsuddeen Hassan  and
      Ekbal, Asif  and
      Das, Uday Kumar",
    booktitle = "Proceedings of the Workshop on Beyond English: Natural Language Processing for all Languages in an Era of Large Language Models",
    month = sep,
    year = "2025",
    address = "Varna, Bulgaria",
    publisher = "INCOMA Ltd., Shoumen, BULGARIA",
    url = "https://preview.aclanthology.org/corrections-2026-01/2025.globalnlp-1.18/",
    pages = "159--171",
    abstract = "Large language models (LLMs) excel at multilingual tasks, yet their internal language processing remains poorly understood. We analyze how Aya-23-8B, a decoder-only LLM trained on balanced multilingual data, handles code-mixed, cloze, and translation tasks compared to predominantly monolingual models like Llama 3 and Chinese-LLaMA-2. Using logit lens and neuron specialization analyses, we find: (1) Aya-23 activates typologically related language representations during translation, unlike English-centric models that rely on a single pivot language; (2) code-mixed neuron activation patterns vary with mixing rates and are shaped more by the base language than the mixed-in one; and (3) Aya-23{'}s language-specific neurons for code-mixed inputs concentrate in final layers, diverging from prior findings on decoder-only models. Neuron overlap analysis further shows that script similarity and typological relations impact processing across model types. These findings reveal how multilingual training shapes LLM internals and inform future cross-lingual transfer research. The code and dataset are publicly available."
}