Reproduce “The Zeno's Paradox of `Low-Resource' Languages” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Hellina Hailu Nigatu, Atnafu Lambebo Tonja, Benjamin Rosman, Thamar Solorio, Monojit Choudhury
This study qualitatively analyzes how 'low-resource languages' are defined in natural language processing literature, revealing that their status is a multi-dimensional construct beyond mere data quantity. It identifies four interconnected aspects—socio-political, resource availability, linguistic artifacts, and community agency—that collectively contribute to a language's 'low-resourcedness', informing more precise measurement and targeted interventions.
Key insights
- A language's 'low-resourcedness' is not a singular concept but a complex, multi-faceted construct influenced by socio-political factors, human and digital resource availability, linguistic artifacts, and community agency.
- Languages categorized within the same data availability tier can exhibit vastly different needs and characteristics across other resource aspects, rendering a purely data-centric definition insufficient.
- There are direct causal links where socio-political issues constrain resource availability, which in turn impacts the production of linguistic artifacts and the agency of community members in technology development.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.