analiza aiqbench
This research addresses the critical bottleneck in machine translation where standard metrics fail to capture nuanced meaning loss, proposing a new ontology for untranslatability. By introducing specific compensation strategies like explanatory context (Annotation), the study provides actionable data for training models that handle semantic gaps rather than forcing one-to-one equivalence. This work advances NLP by shifting focus from raw accuracy benchmarks to preserving cultural and linguistic depth in cross-lingual communication.
Translating the Untranslatable: An Operationalizable Ontology for Untranslatability
arXiv:2606.17354v1 Announce Type: cross Abstract: Untranslatability, cases where meaning cannot be directly preserved across languages, is well-studied in linguistics but underexplored in NLP. As machine translation (MT) systems improve on standard benchmarks, their limitations increasingly concentrate in such cases, where translation cannot be reduced to one-to-one equivalence. We introduce a structured ontology of untranslatability along with a taxonomy of compensation strategies, which are specific techniques to convey meaning under these untranslatable circumstances. We operationalize this framework into a multilingual dataset of untranslatable sentences paired with strategy-based translations, enabling controlled analysis of translation behavior. Initial human preference studies suggest that translation quality depends on the strategy used, with consistent preferences for outputs that include explanatory context, known as the Annotation compensation strategy. Our framework and dataset provide a foundation for studying and modeling strategy-informed machine translation.
Abstract:Untranslatability, cases where meaning cannot be directly preserved across languages, is well-studied in linguistics but underexplored in NLP. As machine translation (MT) systems improve on standard benchmarks, their limitations increasingly concentrate in such cases, where translation cannot be reduced to one-to-one equivalence. We introduce a structured ontology of untranslatability along with a taxonomy of compensation strategies, which are specific techniques to convey meaning under these untranslatable circumstances. We operationalize this framework into a multilingual dataset of untranslatable sentences paired with strategy-based translations, enabling controlled analysis of translation behavior. Initial human preference studies suggest that translation quality depends on the strategy used, with consistent preferences for outputs that include explanatory context, known as the Annotation compensation strategy. Our framework and dataset provide a foundation for studying and modeling strategy-informed machine translation.
Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
arXiv:2606.17354 [cs.CL]
(or arXiv:2606.17354v1 [cs.CL] for this version)
https://doi.org/10.48550/arXiv.2606.17354 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 228 cuvinte