DGS Corpus Sign Language Multi-Tier Annotation

# DGS Corpus Sign Language Multi-Tier Annotation Configuration
# Based on Hanke et al., LREC 2020
# Paper: https://aclanthology.org/2020.lrec-1.720/
# Task: ELAN-style multi-tier annotation of German Sign Language corpus videos

annotation_task_name: "DGS Corpus Sign Language Multi-Tier Annotation"
task_dir: "."

# Data configuration
data_files:
  - sample-data.json
item_properties:
  id_key: "id"
  text_key: "video_url"

# Output
output_annotation_dir: "annotation_output/"
output_annotation_format: "json"

# Annotation schemes - ELAN-style parallel tiers aligned to the video timeline
annotation_schemes:
  # Tier 1: Sign type segmentation
  - name: "sign_type_tier"
    description: |
      Segment the signing stream into individual sign tokens and classify each
      by its linguistic type. Mark precise onset and offset boundaries for
      every manual sign produced by the signer.
    annotation_type: "video_annotation"
    mode: "segment"
    labels:
      - name: "lexical"
        color: "#3B82F6"
        tooltip: "Standard lexical sign from the DGS lexicon"
      - name: "pointing"
        color: "#F59E0B"
        tooltip: "Indexical pointing sign (pronouns, locatives, determiners)"
      - name: "gesture"
        color: "#8B5CF6"
        tooltip: "Non-lexical communicative gesture"
      - name: "classifier"
        color: "#10B981"
        tooltip: "Classifier predicate or depicting sign"
      - name: "fingerspelling"
        color: "#EF4444"
        tooltip: "Manual alphabet spelling of a German word"
      - name: "fragment"
        color: "#6B7280"
        tooltip: "Incomplete or aborted sign"
      - name: "palm-up"
        color: "#EC4899"
        tooltip: "Palm-up gesture (discourse marker or uncertainty)"
    show_timecode: true
    video_fps: 25

  # Tier 2: Mouth gesture patterns
  - name: "mouth_gesture_tier"
    description: |
      Annotate the mouth activity accompanying each sign. Distinguish between
      mouth gestures (language-internal, non-derived from spoken language),
      mouthings (derived from German words), and neutral mouth positions.
    annotation_type: "video_annotation"
    mode: "segment"
    labels:
      - name: "mouth-gesture"
        color: "#06B6D4"
        tooltip: "Language-internal mouth gesture not derived from spoken German"
      - name: "mouthing"
        color: "#84CC16"
        tooltip: "Mouth pattern derived from a German spoken word"
      - name: "mouth-open"
        color: "#F97316"
        tooltip: "Open mouth without clear gestural or mouthing pattern"
      - name: "neutral"
        color: "#9CA3AF"
        tooltip: "Neutral or resting mouth position"
    show_timecode: true
    video_fps: 25

  # Tier 3: Non-manual signals
  - name: "nonmanual_tier"
    description: |
      Annotate non-manual signals that carry grammatical or pragmatic meaning.
      These include head movements, eyebrow activity, eye gaze shifts, and
      body movements that co-occur with manual signing.
    annotation_type: "video_annotation"
    mode: "segment"
    labels:
      - name: "head-nod"
        color: "#22C55E"
        tooltip: "Vertical head nod (affirmation, agreement, assertion)"
      - name: "head-shake"
        color: "#EF4444"
        tooltip: "Horizontal head shake (negation, disagreement)"
      - name: "brow-raise"
        color: "#A855F7"
        tooltip: "Raised eyebrows (yes/no question, topic marker, conditional)"
      - name: "brow-furrow"
        color: "#6366F1"
        tooltip: "Furrowed eyebrows (wh-question, focus, intensity)"
      - name: "eye-gaze-shift"
        color: "#14B8A6"
        tooltip: "Deliberate shift in eye gaze direction (role shift, reference)"
      - name: "squint"
        color: "#F59E0B"
        tooltip: "Narrowed eyes (specificity, shared knowledge, rhetorical question)"
      - name: "shoulder-shift"
        color: "#EC4899"
        tooltip: "Lateral shoulder shift (role shift, contrastive topics)"
    show_timecode: true
    video_fps: 25

  # Tier 4: Discourse function classification
  - name: "discourse_function"
    description: "Classify the discourse function of the current utterance or clause."
    annotation_type: radio
    labels:
      - "statement"
      - "question"
      - "negation"
      - "topic"
      - "comment"
      - "emphasis"
    keyboard_shortcuts:
      statement: "1"
      question: "2"
      negation: "3"
      topic: "4"
      comment: "5"
      emphasis: "6"

  # Tier 5: German translation (free text)
  - name: "german_translation"
    description: "Provide a fluent German translation of the signed content in this clip."
    annotation_type: text
    textarea: true

# HTML layout
html_layout: |
  <div style="max-width: 900px; margin: 0 auto;">
    <h3 style="margin-bottom: 8px;">DGS Corpus: Multi-Tier German Sign Language Annotation</h3>
    <p style="color: #666; font-size: 14px; margin-bottom: 16px;">
      Annotate German Sign Language (DGS) videos across multiple parallel tiers,
      following ELAN-style conventions used in the DGS Corpus project.
    </p>
    <div style="text-align: center; margin-bottom: 20px;">
      <video controls width="720" style="max-width: 100%; border-radius: 8px; border: 1px solid #ddd;">
        <source src="{{video_url}}" type="video/mp4">
        Your browser does not support video playback.
      </video>
    </div>
    <div style="background: #f8f9fa; padding: 12px; border-radius: 6px; margin-bottom: 16px; font-size: 13px;">
      <strong>Multi-Tier Instructions:</strong> Annotate across five parallel tiers: sign types,
      mouth gestures, non-manual signals, discourse function, and German translation. Each tier
      captures a different linguistic layer of the DGS utterance.
    </div>
  </div>

# User configuration
allow_all_users: true

# Task assignment
instances_per_annotator: 30
annotation_per_instance: 2

# Instructions
annotation_instructions: |
  ## DGS Corpus Multi-Tier Annotation

  This task applies ELAN-style multi-tier annotation to German Sign Language (DGS)
  videos from the Public DGS Corpus.

  ### Tier 1: Sign Type Segmentation
  - Segment the signing stream into individual tokens
  - Classify each sign by type:
    - **Lexical**: Standard DGS signs from the lexicon
    - **Pointing**: Index signs for pronouns, locations, determiners
    - **Gesture**: Non-lexical communicative gestures
    - **Classifier**: Depicting signs and classifier predicates
    - **Fingerspelling**: Manual alphabet spellings of German words
    - **Fragment**: Incomplete or self-corrected signs
    - **Palm-up**: Palm-up gestures used as discourse markers

  ### Tier 2: Mouth Activity
  - Annotate mouth patterns in parallel with manual signs:
    - **Mouth gesture**: DGS-internal mouth component (not from spoken German)
    - **Mouthing**: Mouth pattern derived from a German spoken word
    - **Mouth open**: Open mouth without clear linguistic function
    - **Neutral**: Resting or closed mouth position

  ### Tier 3: Non-Manual Signals
  - Mark grammatically significant non-manual signals:
    - **Head nod/shake**: Affirmation vs. negation
    - **Brow raise**: Yes/no questions, topics, conditionals
    - **Brow furrow**: Wh-questions, focus, intensification
    - **Eye gaze shift**: Role shift, referent tracking
    - **Squint**: Specificity, shared knowledge
    - **Shoulder shift**: Role shift, contrastive topics

  ### Tier 4: Discourse Function
  - Classify the overall discourse function of the utterance

  ### Tier 5: German Translation
  - Provide a natural German translation of the signed content

  ### Quality Notes
  - Pay special attention to non-manual markers that change sentence type
  - Mouth gestures and mouthings may overlap with sign boundaries differently
  - Use frame-by-frame navigation for precise boundary placement
  - For dialogues, annotate only the designated signer's productions