CHILDES Child Language Multi-Tier Annotation

# CHILDES Child Language Multi-Tier Annotation Configuration
# Based on MacWhinney, Journal of Child Language 2000
# Paper: https://doi.org/10.1017/S0305000900003581
# Task: ELAN-style multi-tier annotation of child-adult interaction for language acquisition

annotation_task_name: "CHILDES Child 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: Utterance boundary segmentation
  - name: "utterance_tier"
    description: |
      Segment the video timeline into individual utterances. Mark who is
      speaking (child or adult) and identify overlapping speech, non-verbal
      vocalizations, and unintelligible segments.
    annotation_type: "video_annotation"
    mode: "segment"
    labels:
      - name: "child-utterance"
        color: "#3B82F6"
        tooltip: "Clear, interpretable utterance produced by the child"
      - name: "adult-utterance"
        color: "#10B981"
        tooltip: "Utterance produced by the adult caregiver or interlocutor"
      - name: "overlap"
        color: "#F59E0B"
        tooltip: "Child and adult speaking simultaneously"
      - name: "vocalization"
        color: "#A855F7"
        tooltip: "Non-linguistic vocalization (babbling, cooing, crying, laughing)"
      - name: "unintelligible"
        color: "#EF4444"
        tooltip: "Speech that cannot be reliably transcribed"
    show_timecode: true
    video_fps: 25

  # Tier 2: Morphological annotation (free text)
  - name: "morphology"
    description: |
      Provide a morphological annotation of the child's utterance using CHAT
      conventions. Break words into morphemes and mark inflectional morphology
      (e.g., want-3SG, go-PAST, dog-PL).
    annotation_type: text
    textarea: true

  # Tier 3: Syntactic annotation (free text)
  - name: "syntax"
    description: |
      Provide a syntactic annotation of the child's utterance. Note phrase
      structure, word order patterns, and any syntactic constructions
      (e.g., SVO, Wh-question, negation, relative clause).
    annotation_type: text
    textarea: true

  # Tier 4: Communicative context classification
  - name: "communicative_context"
    description: "Classify the communicative context or function of the child's utterance."
    annotation_type: radio
    labels:
      - "spontaneous"
      - "imitation"
      - "routine"
      - "response"
      - "question"
      - "self-talk"
      - "directed-speech"
    keyboard_shortcuts:
      spontaneous: "1"
      imitation: "2"
      routine: "3"
      response: "4"
      question: "5"
      self-talk: "6"
      directed-speech: "7"

  # Tier 5: Error type classification
  - name: "error_type"
    description: "Classify the type of linguistic error in the child's utterance, if any."
    annotation_type: radio
    labels:
      - "none"
      - "phonological"
      - "morphological"
      - "syntactic"
      - "lexical"
      - "pragmatic"
    keyboard_shortcuts:
      none: "q"
      phonological: "w"
      morphological: "e"
      syntactic: "r"
      lexical: "t"
      pragmatic: "y"

# HTML layout
html_layout: |
  <div style="max-width: 900px; margin: 0 auto;">
    <h3 style="margin-bottom: 8px;">CHILDES: Multi-Tier Child Language Annotation</h3>
    <p style="color: #666; font-size: 14px; margin-bottom: 16px;">
      Annotate child-adult interaction videos across multiple tiers for utterance
      boundaries, morphology, syntax, communicative context, and error analysis.
    </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 the child-adult interaction across
      five parallel tiers: utterance segmentation, morphological coding, syntactic structure,
      communicative context, and error classification. Focus primarily on the child's productions.
    </div>
  </div>

# User configuration
allow_all_users: true

# Task assignment
instances_per_annotator: 30
annotation_per_instance: 2

# Instructions
annotation_instructions: |
  ## CHILDES Child Language Multi-Tier Annotation

  This task uses ELAN-style multi-tier annotation for child-adult interaction
  videos following CHILDES/TalkBank conventions.

  ### Tier 1: Utterance Boundary Segmentation
  - Segment the timeline into individual utterances:
    - **Child utterance**: Interpretable speech produced by the child
    - **Adult utterance**: Speech from the caregiver or adult interlocutor
    - **Overlap**: Simultaneous speech from both parties
    - **Vocalization**: Non-linguistic sounds (babbling, cooing, crying, laughing)
    - **Unintelligible**: Speech that cannot be reliably understood
  - Use intonation contours and pauses to determine utterance boundaries
  - An utterance is defined as a single communicative unit with one intonation contour

  ### Tier 2: Morphological Annotation
  - Code the child's utterance morphologically using CHAT-style notation:
    - Separate morphemes with hyphens: "want-3SG", "dog-PL", "go-PAST"
    - Mark overregularizations: "go-ed" for "went" (overregularized past)
    - Use standard glosses: PL (plural), PAST (past tense), PROG (progressive),
      3SG (third person singular), POSS (possessive), NEG (negation)
  - Leave blank for adult utterances or unintelligible segments

  ### Tier 3: Syntactic Annotation
  - Note the syntactic structure of the child's utterance:
    - Word order pattern (SVO, SV, VO, single word, etc.)
    - Sentence type (declarative, interrogative, imperative, exclamatory)
    - Notable constructions (negation, questions, relative clauses, coordination)
    - Missing obligatory elements (e.g., "want cookie" = missing determiner)

  ### Tier 4: Communicative Context
  - Classify the communicative function:
    - **Spontaneous**: Self-initiated utterance, not prompted
    - **Imitation**: Direct repetition of an adult model
    - **Routine**: Part of a practiced routine (counting, song, greeting)
    - **Response**: Answer to an adult question or prompt
    - **Question**: Child asking a question
    - **Self-talk**: Speech directed to self or toys, not to the adult
    - **Directed speech**: Speech clearly addressed to a specific person

  ### Tier 5: Error Classification
  - Identify the primary error type, if any:
    - **None**: Target-like production
    - **Phonological**: Sound substitution, deletion, or addition
    - **Morphological**: Missing or incorrect inflection (e.g., "goed" for "went")
    - **Syntactic**: Word order error, missing function words
    - **Lexical**: Wrong word choice or neologism
    - **Pragmatic**: Contextually inappropriate utterance

  ### Developmental Notes
  - Child age is provided in the metadata; keep developmental expectations in mind
  - What counts as an "error" depends on the child's age and stage
  - At early stages (12-24 months), single words and babbling are expected
  - By 36+ months, expect more complex multi-word utterances