State Space Models are Effective Sign Language Learners: Exploiting Phonological Compositionality for Vocabulary-Scale Recognition

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cheng, Bryan, Jin, Austin, Zhang, Jasper
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913020840509440
author Cheng, Bryan
Jin, Austin
Zhang, Jasper
author_facet Cheng, Bryan
Jin, Austin
Zhang, Jasper
contents Sign language recognition suffers from catastrophic scaling failure: models achieving high accuracy on small vocabularies collapse at realistic sizes. Existing architectures treat signs as atomic visual patterns, learning flat representations that cannot exploit the compositional structure of sign languages-systematically organized from discrete phonological parameters (handshape, location, movement, orientation) reused across the vocabulary. We introduce PHONSSM, enforcing phonological decomposition through anatomically-grounded graph attention, explicit factorization into orthogonal subspaces, and prototypical classification enabling few-shot transfer. Using skeleton data alone on the largest ASL dataset ever assembled (5,565 signs), PHONSSM achieves 72.1% on WLASL2000 (+18.4pp over skeleton SOTA), surpassing most RGB methods without video input. Gains are most dramatic in the few-shot regime (+225% relative), and the model transfers zero-shot to ASL Citizen, exceeding supervised RGB baselines. The vocabulary scaling bottleneck is fundamentally a representation learning problem, solvable through compositional inductive biases mirroring linguistic structure.
format Preprint
id arxiv_https___arxiv_org_abs_2604_08761
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle State Space Models are Effective Sign Language Learners: Exploiting Phonological Compositionality for Vocabulary-Scale Recognition
Cheng, Bryan
Jin, Austin
Zhang, Jasper
Computer Vision and Pattern Recognition
Sign language recognition suffers from catastrophic scaling failure: models achieving high accuracy on small vocabularies collapse at realistic sizes. Existing architectures treat signs as atomic visual patterns, learning flat representations that cannot exploit the compositional structure of sign languages-systematically organized from discrete phonological parameters (handshape, location, movement, orientation) reused across the vocabulary. We introduce PHONSSM, enforcing phonological decomposition through anatomically-grounded graph attention, explicit factorization into orthogonal subspaces, and prototypical classification enabling few-shot transfer. Using skeleton data alone on the largest ASL dataset ever assembled (5,565 signs), PHONSSM achieves 72.1% on WLASL2000 (+18.4pp over skeleton SOTA), surpassing most RGB methods without video input. Gains are most dramatic in the few-shot regime (+225% relative), and the model transfers zero-shot to ASL Citizen, exceeding supervised RGB baselines. The vocabulary scaling bottleneck is fundamentally a representation learning problem, solvable through compositional inductive biases mirroring linguistic structure.
title State Space Models are Effective Sign Language Learners: Exploiting Phonological Compositionality for Vocabulary-Scale Recognition
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2604.08761