Grounding or Guessing? Visual Signals for Detecting Hallucinations in Sign Language Translation

TL;DR

This paper introduces a reliability measure based on visual grounding signals to detect hallucinations in sign language translation models, improving robustness and interpretability.

Contribution

It proposes a novel token-level reliability measure combining feature sensitivity and counterfactual signals to identify hallucinations in SLT models.

Findings

Reliability predicts hallucination rates across datasets and models.

Reliability decreases with visual degradation, indicating sensitivity to visual grounding.

Combining reliability with text signals enhances hallucination risk estimation.

Abstract

Hallucination, where models generate fluent text unsupported by visual evidence, remains a major flaw in vision-language models and is particularly critical in sign language translation (SLT). In SLT, meaning depends on precise grounding in video, and gloss-free models are especially vulnerable because they map continuous signer movements directly into natural language without intermediate gloss supervision that serves as alignment. We argue that hallucinations arise when models rely on language priors rather than visual input. To capture this, we propose a token-level reliability measure that quantifies how much the decoder uses visual information. Our method combines feature-based sensitivity, which measures internal changes when video is masked, with counterfactual signals, which capture probability differences between clean and altered video inputs. These signals are aggregated into a sentence-level reliability score, providing a compact and interpretable measure of visual grounding. We evaluate the proposed measure on two SLT benchmarks (PHOENIX-2014T and CSL-Daily) with both gloss-based and gloss-free models. Our results show that reliability predicts hallucination rates, generalizes across datasets and architectures, and decreases under visual degradations. Beyond these quantitative trends, we also find that reliability distinguishes grounded tokens from guessed ones, allowing risk estimation without references; when combined with text-based signals (confidence, perplexity, or entropy), it further improves hallucination risk estimation. Qualitative analysis highlights why gloss-free models are more susceptible to hallucinations. Taken together, our findings establish reliability as a practical and reusable tool for diagnosing hallucinations in SLT, and lay the groundwork for more robust hallucination detection in multimodal generation.