Get Started
Research questionWhat language should a language model produce when its training data are translated, synthetic, or heavily curated?Translated, synthetic, and curated corpora can shape a model's outputs around artifacts of data production rather than a clearly defined language. This makes it difficult to say what language the model represents or should produce.
AI
Natural Language Processing
Research Paper
Latest papersRecent research connected to this question, newest first.Language, Language Models, and What We're Talking AboutThe discussion uses Italian language models trained or specialised on translated and synthetic data, with evaluation on similarly unnatural data, as evidence for examining these distinctions.research paper · Sep 3, 2026
Related questions
How can models adapt to low-resource languages without damaging source-language and related-task performance?How can we test whether language models causally use scientific mechanisms instead of answer-correlated shortcuts?Can language-based models replace specialized architectures for structured data without sacrificing structural representation and computation?How can scientific agents choose domain-specific procedures that make analyses defensible?
Home
Topics
Search
Library