---
title: "DiscoSign: Discourse-Aware Text to Sign Language Gloss Translation"
description: "DiscoSign addresses a limitation in sign language processing: systems have traditionally translated at the sentence level, overlooking discourse phenomena important to sign language comprehension. It presents a modular Large Language Model (LLM)-based framework for text-to-sign language gloss translation that handles spatial coreference resolution, Question-Answer Clauses (QACs), and concept-gloss consistency, preserving entity locations and stable English-to-ASL sign mappings across discourse. Because traditional translation metrics do not measure discourse-level quality, the authors introduce metrics for each coherence dimension targeted by the framework. Experiments on sentence-level and discourse-level datasets report improved spatial consistency and entity tracking over sentence-only translation, while maintaining competitive single-sentence gloss quality; the work describes this as the first systematic framework with a corresponding evaluation methodology."
---

# DiscoSign: Discourse-Aware Text to Sign Language Gloss Translation

[Apple](https://yomu.fyi/company/apple) · Sep 11, 2026

**Type:** Problem & solution

## Summary

DiscoSign addresses a limitation in sign language processing: systems have traditionally translated at the sentence level, overlooking discourse phenomena important to sign language comprehension. It presents a modular Large Language Model (LLM)-based framework for text-to-sign language gloss translation that handles spatial coreference resolution, Question-Answer Clauses (QACs), and concept-gloss consistency, preserving entity locations and stable English-to-ASL sign mappings across discourse. Because traditional translation metrics do not measure discourse-level quality, the authors introduce metrics for each coherence dimension targeted by the framework. Experiments on sentence-level and discourse-level datasets report improved spatial consistency and entity tracking over sentence-only translation, while maintaining competitive single-sentence gloss quality; the work describes this as the first systematic framework with a corresponding evaluation methodology.

## Context

Sign language processing systems have traditionally operated at the sentence level, ignoring discourse phenomena that are fundamental to sign language comprehension. Traditional translation metrics also fail to capture discourse-level quality.

## Approach / What changed

DiscoSign uses a modular Large Language Model (LLM)-based translation framework to address spatial coreference resolution, Question-Answer Clauses (QACs), and concept-gloss consistency. It also introduces evaluation metrics for the discourse-coherence dimensions addressed by the framework.

## Takeaways

- The framework addresses spatial coreference, QACs, and concept-gloss consistency in text-to-ASL gloss translation.
- The evaluation suite is designed to measure discourse coherence dimensions that traditional translation metrics do not capture.
- Experiments report improved spatial consistency and entity tracking over sentence-only translation while retaining competitive single-sentence gloss translation quality.

**Tags:** [ASL](https://yomu.fyi/topic/asl), [LLMs](https://yomu.fyi/topic/llm)

- Source: [Apple](https://machinelearning.apple.com/research/discosign-gloss-translation)
- Source URL: https://machinelearning.apple.com/research/discosign-gloss-translation
- Ingested by Yomu: 2026-09-11T16:03:48.316Z

[Read original post](https://machinelearning.apple.com/research/discosign-gloss-translation)