@inproceedings{mitra-etal-2026-dual,
    title = "Dual-Axis Compositional Contrastive Few-Shot Learning using Prototypes Across Linguistic and Semantic Dimensions for {I}ndic Low-Resource Multilingual {NLU}",
    author = "Mitra, Kathakali  and
      Singh, Sakshi  and
      Gunapati, Sree Nithish Reddy  and
      Malapati, Aruna  and
      Lee, Mark G.",
    editor = "Chakravarthi, Bharathi Raja  and
      B, Bharathi  and
      Buitelaar, Paul  and
      Thenmozhi, Durairaj  and
      Garc{\'i}a Cumbreras, Miguel {\'A}ngel  and
      Jim{\'e}nez Zafra, Salud Mar{\'i}a",
    booktitle = "Proceedings of the Sixth Workshop on Language Technology for Equality, Diversity, Inclusion",
    month = jul,
    year = "2026",
    address = "Virtual (Online)",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl-workshops/2026.ltedi-1.3/",
    pages = "27--36",
    ISBN = "979-8-89176-424-8",
    abstract = "Multilingual Natural Language Understanding (NLU) systems often struggle to adapt when new languages or new semantic labels are introduced with only a few annotated examples. This challenge is particularly pronounced for low-resource languages, where limited supervision and evolving label spaces make conventional joint-label classification approaches unstable. Most existing multilingual NLU models treat each language-semantic pair as an independent class, entangling linguistic and semantic representations and hindering few-shot adaptation. We propose Dual-Axis Compositional Few-Shot Learning, a framework that explicitly factorizes the representation space into linguistic and semantic embedding axes, enabling independent modeling of language variation and domain-intent semantics. Joint representations are constructed compositionally through multiplicative interaction of axis-specific embeddings, allowing controlled adaptation when either the language set or the semantic label space evolves. The framework integrates factorized prototype learning, axis-structured contrastive alignment, and disentanglement regularization using HSIC-based statistical independence and Jacobian-based cross-axis decorrelation. Experiments on six low-resource Indic languages spanning Indo-Aryan and Dravidian families (Hindi, Bengali, Sanskrit, Assamese, Tamil, and Telugu) demonstrate strong performance under two structured generalization regimes. The model achieves 81.12{\%} accuracy when adapting to few-shot languages with known semantics and 63.5{\%} accuracy when learning new semantic classes from few-shot examples, along with an accuracy of 89.56{\%} on known language and seen semantics. These results show that axis-factorized representations enable stable compositional generalization, offering a promising direction for scalable multilingual NLU in linguistically diverse low-resource settings."
}