PSL SignBank: A Multimodal Machine Readable Dictionary for Pakistan Sign Language

Document Type

Article

Publication Date

1-1-2026

Abstract

The deaf community in Pakistan faces significant communication barriers due to the absence of standardized, machine-readable resources for Pakistani Sign Language (PSL). To address this challenge, SignBank for Pakistani Sign Language (PSL SignBank) has been developed as a machine-readable dictionary to preserve and promote PSL. The corpus includes 300 commonly used English words. Each word is translated into Urdu and encoded using HamNoSys notation languageindependent phonetic transcription system. Each entry of dictionary integrate multiple modalities including English word, Urdu translation, HamNoSys vector representation, human signer video, and avatar-generated animation via SiGML (Signing Gesture Markup Language) rendering. The development of corpus involved systematic video recording with deaf participants from multiple institutions. This procedure was followed by the team of three sign language experts and two interpreters who verified gestural accuracy in the shape, movement, and location parameters of the hand. Compared to traditional video-based dictionaries, PSL SignBank achieved approximately 95% storage reduction with HamNoSys notation requiring around 1 KB per sign versus 1 MB for video and supports scalable sentence-level translation through the concatenation of machine-readable notations. The avatar-based rendering system was validated against human signer videos which confirmed accurate gesture reproduction for both static and dynamic signs. This work establish a foundational infrastructure for computational PSL applications that include text-to-sign translation systems, sign language recognition models, and educational platforms. PSL SignBank represents a critical advance towards accessibility, digital inclusion, and empowerment of Pakistan’s deaf community and it also provide a replicable framework for under-resourced sign language documentation globally.

Keywords

Multimodal Corpus, PSL vocabulary, SigML tags, Sign Language Dictionary

Publication Title

Vfast Transactions on Software Engineering

ISSN

24116246

DOI

10.21015/vtse.v14i1.2246

Volume

14

Issue

1

First Page

206

Last Page

224

Share

COinS