UD-English-CHILDES: A Collected Resource of Gold and Silver Universal Dependencies Trees for Child Language Interactions

2025-04-28

AI Summary

Existing CHILDES corpora suffer from annotation heterogeneity and limited scale, hindering standardized dependency parsing and cross-linguistic research in child language acquisition. Method: We propose a “gold + silver” collaborative annotation paradigm: (1) manually constructing 48k high-quality Universal Dependencies (UD) v2-compliant dependency trees across 11 CHILDES subcorpora, covering both child and caregiver utterances; and (2) automatically generating 1M silver-standard annotations via rule-based and model-based approaches. We further perform transcription alignment, noise cleaning, and cross-speaker/corpus dependency consistency verification. Contribution/Results: This work delivers the first officially released CHILDES-driven UD treebank—uniquely standardized under UD v2 for child language. It enables systematic UD adoption and large-scale expansion in this domain, significantly enhancing data support for child language parsing, acquisition modeling, and cross-lingual dependency transfer research.

Technology Category

• Natural Language Processing: Syntax — Tagging, Chunking & Parsing
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

CHILDES is a widely used resource of transcribed child and child-directed speech. This paper introduces UD-English-CHILDES, the first officially released Universal Dependencies (UD) treebank derived from previously dependency-annotated CHILDES data with consistent and unified annotation guidelines. Our corpus harmonizes annotations from 11 children and their caregivers, totaling over 48k sentences. We validate existing gold-standard annotations under the UD v2 framework and provide an additional 1M silver-standard sentences, offering a consistent resource for computational and linguistic research.

Problem

• Creating first UD treebank from CHILDES data
• Harmonizing annotations for child-caregiver interactions
• Providing gold and silver standard UD resources

Innovation

• First UD treebank from CHILDES data
• Harmonized annotations from 11 children
• Validated gold and silver UD annotations