Motion Engine
The Motion Engine is the future movement layer for ViaSign.
It is intended to translate structured grammar information into motion-ready signing instructions that can later support avatar output, review, and auditing.
The Motion Engine does not replace the parser.
It builds on top of it.
Parser V2 helps identify sentence structure, grammar signals, review needs, and motion-relevant information.
The Motion Engine explores how those signals could become movement.
Why a Motion Engine Exists
Sign language output is not only about choosing the right signs.
It also involves how signs move, where they happen, how they transition, and how meaning is carried through space, expression, timing, and body movement.
A simple word-to-sign system is not enough.
For signed languages, motion matters because meaning can depend on:
- handshape
- palm orientation
- movement path
- location
- direction
- facial expression
- body posture
- timing
- signing space
- transitions between signs
The Motion Engine exists to help make this movement layer more structured and auditable.
Grammar Before Motion
The Motion Engine follows the same principle as Parser V2:
Grammar before motion.
This means motion should not be generated blindly from English words.
The system should first understand the sentence structure.
Then it can decide what parts may need motion, space, direction, timing, or review.
For example:
- a question may need facial expression or timing markers
- negation may need non-manual signals or specific placement
- a time marker may need to be established before the main sentence
- a person or object reference may need signing space
- a directional action may need motion from one reference point to another
This is why the Motion Engine depends on structured parser output.
Motion Should Be Auditable
The goal is not only to make an avatar move.
The goal is to make movement understandable and reviewable.
An auditable motion system should make it possible to ask:
- What motion was generated?
- Why was that motion chosen?
- Which parser signal triggered it?
- Does this motion need review?
- Is this movement final or experimental?
- Which parts are grammar-driven?
- Which parts are placeholder motion?
This matters because polished animation can create false confidence.
A movement may look smooth but still be linguistically wrong.
For ViaSign, motion should be inspectable before it becomes presentation.
Possible Motion Layers
The Motion Engine may include several layers over time.
Hand Pose Layer
This layer describes the shape and configuration of the hands.
It may include:
- finger curl
- thumb position
- handshape tokens
- palm orientation
- dominant and non-dominant hand roles
Movement Layer
This layer describes how the hands move.
It may include:
- path
- direction
- repetition
- speed
- hold
- transition
- start and end positions
Signing Space Layer
This layer describes where signs happen in relation to the signer.
It may include:
- neutral space
- left and right reference points
- object placement
- person indexing
- directional relationships
Non-Manual Signal Layer
This layer describes meaning carried outside the hands.
It may include:
- facial expression
- eyebrow position
- head movement
- body shift
- gaze
- mouth pattern placeholders
This layer is especially important because signed languages are not hand-only languages.
Review Layer
This layer marks which motion decisions should be reviewed.
It may include:
- uncertain movement
- placeholder motion
- grammar-sensitive output
- culturally sensitive output
- motion that requires review led by Deaf community members
What the Motion Engine Is Not
The Motion Engine is not a finished avatar system.
It is not a complete Singapore Sign Language motion generator.
It is not a replacement for Deaf signers, teachers, interpreters, or reviewers.
It is not a claim that AI-generated movement is automatically correct.
It is an engineering layer for exploring how signing motion can be represented, reviewed, and improved over time.
Relationship to Parser V2
Parser V2 and the Motion Engine should work together.
A possible flow is:
- Input sentence
- Parser V2 analysis
- Grammar and structure signals
- Review flags
- Motion-ready information
- Motion Engine planning
- Motion tokens or movement instructions
- Human or community review
- Future avatar output
This keeps the system layered.
Each layer can be tested separately.
Each layer can expose uncertainty.
Each layer can be improved without pretending the whole system is complete.
Review-First Motion
For ViaSign, motion should be review-first.
This means the system should not only produce movement.
It should also explain what it produced and where review is needed.
A review-first motion system should make space for correction, feedback, and community guidance.
This is important because signed languages are living languages.
They are shaped by people, culture, usage, and context.
The Motion Engine should respect that by staying inspectable, cautious, and open to review.
Guiding Principle
The Motion Engine exists because sign language accessibility cannot stop at text.
But it also cannot jump straight from text into animation without care.
The goal is to build a motion layer that is structured, auditable, reviewable, and respectful of signed language complexity.
Not just movement.
Meaningful movement.
Not just animation.
Reviewed motion.
Not just output.
A system that can be checked, corrected, and improved over time.