{"slug": "dualanchor-preserving-language-priors-and-improving-lexical-fidelity-in-gloss", "title": "DualAnchor: Preserving Language Priors and Improving Lexical Fidelity in Gloss-Free Sign Language Translation", "summary": "Researchers propose DualAnchor, a gloss-free sign language translation framework that preserves language priors and improves lexical fidelity, achieving strong performance on PHOENIX-2014T and CSL-Daily. The framework uses Token-level Prior Anchoring (TPA) to regularize the multimodal decoder toward a frozen LLM's next-token distribution, and Optimal Transport Alignment (OTA) to align visual and textual tokens via Sinkhorn optimization. The study attributes fluency gains to TPA and reduced lexical errors to OTA.", "body_md": "arXiv:2607.27614v1 Announce Type: new\nAbstract: Recent advances in large language models (LLMs) have led sign language translation (SLT), the task of converting sign-language videos into spoken-language text, to increasingly adopt LLMs as textual backbones. However, despite their strong language modeling capabilities, existing LLM-based SLT methods often undermine rather than exploit this language prior, producing disfluent translations, a failure we term language-prior degradation. Meanwhile, existing methods typically align videos and text at the sentence level, which does not ensure accurate lexical details and creates a lexical fidelity gap. To address both issues, we propose DualAnchor, a gloss-free LLM-based SLT training framework that couples two complementary anchors for linguistically fluent and visually faithful generation. Token-level Prior Anchoring (TPA) preserves the LLM's language prior by regularizing the multimodal decoder at each decoding step toward the next-token distribution of a frozen LLM conditioned on the same autoregressive prefix. Optimal Transport Alignment (OTA) improves lexical fidelity by formulating visual-textual matching as entropy-regularized partial optimal transport, with Sinkhorn optimization inducing a soft alignment between visual tokens and textual content tokens under a cosine cost. DualAnchor achieves strong overall performance on both PHOENIX-2014T and CSL-Daily. Targeted analyses attribute these gains to the complementary effects of the two anchors: TPA improves fluency, whereas OTA reduces fine-grained lexical errors.", "url": "https://wpnews.pro/news/dualanchor-preserving-language-priors-and-improving-lexical-fidelity-in-gloss", "canonical_source": "https://www.machinebrief.com/news/dualanchor-preserving-language-priors-and-improving-lexical-0lmd", "published_at": "2026-07-31 04:00:00+00:00", "updated_at": "2026-07-31 04:37:12.741959+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "natural-language-processing"], "entities": ["DualAnchor", "PHOENIX-2014T", "CSL-Daily"], "alternates": {"html": "https://wpnews.pro/news/dualanchor-preserving-language-priors-and-improving-lexical-fidelity-in-gloss", "markdown": "https://wpnews.pro/news/dualanchor-preserving-language-priors-and-improving-lexical-fidelity-in-gloss.md", "text": "https://wpnews.pro/news/dualanchor-preserving-language-priors-and-improving-lexical-fidelity-in-gloss.txt", "jsonld": "https://wpnews.pro/news/dualanchor-preserving-language-priors-and-improving-lexical-fidelity-in-gloss.jsonld"}}