Language engine
Work toward sign-aware structure with conservative review flags; the current public parser has no SgSL grammar rules.
ViaSign by Viathorne Labs · Research toward future SgSL support
Building toward SgSL support with care.
ViaSign is an early-stage research project exploring future support for Singapore Sign Language. The current public parser exposes safety contracts, closed natural-English surface categories, and visible uncertainty—not SgSL grammar or translation.
Not a replacement for interpreters, teachers, or Deaf knowledge. A careful prototype path toward better access.
Current status
Language engine first.
Public safety contracts and review flags first. SgSL grammar, verified signing, and motion remain future, human-reviewed work.
Early prototype Human review Motion unavailable
Core idea: sign language should be treated as visual language, not just text conversion.
Care before hypePublic parser · Local integration
Try the review-first boundary.
This early tester identifies a few natural-English surface categories while keeping SgSL grammar unavailable. It does not translate, sign, or create motion.
Before you test
- Use a short, non-sensitive sentence.
- Do not enter names, contact details, medical, legal, safety-critical, or emergency information.
- Every result requires human review and remains motion-not-ready.
- Inputs are sent only when you press the button.
- Testing has a shared-network limit. If it is reached, your sentence stays in this field and the page tells you when to try again.
Read the Privacy notice and prototype limits before wider public testing.
The tester is not configured. ViaSign information remains available below.
Public parser result
Review details
- Human review
- Required
- Motion ready
- No
- Surface intent
- Unknown
- Question kind
- Not analyzed
- Question category
- Not analyzed
- Negation cue
- Not analyzed
Warnings and limits
SgSL output
No SgSL form or gloss proposed.
What ViaSign is
A research path toward sign-aware language support.
ViaSign researches how future AI systems might support SgSL access by making language structure and uncertainty visible: sentence patterns, time markers, questions, indexing, review needs, and future motion planning.
The first goal is not to generate perfect signing. The first goal is to build a reliable foundation that can be tested, reviewed, corrected, and improved step by step.
How it grows
Four layers, built in order.
Sign packets
Represent sign output as movement, location, timing, orientation, and non-manual information.
Motion audit
Check whether a sign plan is readable before it becomes avatar or visual output.
Visual output
Explore future signing interfaces only after the language and motion layers become reliable.
Community review
Designed for community review, not authority over it.
ViaSign should grow through feedback from Deaf people, signers, learners, teachers, and reviewers. The system should make uncertainty visible instead of hiding it behind confident AI output.
This means review flags, clear limits, careful wording, and no claim that the system is official or complete before the community has shaped it.
Current work
What is being built now.
Public safety contracts and visible uncertainty
Closed natural-English surface categories with no returned sentence values
Clean-room parser boundary with no public SgSL grammar rules yet
Private research benchmarks kept outside the public API
Review flags when the system is unsure
Motion and avatar work kept outside the public parser
Principles
Accessibility work needs care, not hype.
With, not over
ViaSign is grounded in the founder's Deaf and hard-of-hearing lived experience and should grow with feedback from Deaf people, signers, teachers, and learners.
Support, not replacement
ViaSign is not an interpreter, teacher, or official source. It is an early research direction for future access and learning support.
Review before authority
Early outputs must stay transparent, conservative, and reviewable instead of pretending to be fully correct.
Founder note
Built from lived experience and a gap that could not be ignored.
ViaSign began from the founder's Deaf and hard-of-hearing lived experience and the belief that sign language and Deaf access should not sit at the edge of technology. That experience informs the direction; it does not represent broad Deaf-community validation or endorsement. The project remains early, with room for review, correction, and future community shaping.
Next step
The language engine comes first.
The current public focus is contract safety, visible uncertainty, closed natural-English surface analysis, and independently authored tests. SgSL grammar research, reviewed examples, sign packets, motion, and visual interfaces remain separate or future work.