What Exactly do Children Receive in Language Acquisition? A Case Study on CHILDES with Automated Detection of Filler-Gap Dependencies

TL;DR

This paper introduces a system that automatically detects filler-gap dependencies in child-directed speech, enabling detailed analysis of children's language input and development trajectories, with implications for understanding language acquisition.

Contribution

The authors develop a novel parsing-based system to identify filler-gap constructions and extraction sites in spoken English corpora, validated on annotated data and applied to CHILDES datasets.

Findings

System achieves high accuracy on annotated data

Characterizes children's filler-gap input over development

Demonstrates utility with a case study on language models

Abstract

Children's acquisition of filler-gap dependencies has been argued by some to depend on innate grammatical knowledge, while others suggest that the distributional evidence available in child-directed speech suffices. Unfortunately, the relevant input is difficult to quantify at scale with fine granularity, making this question difficult to resolve. We present a system that identifies three core filler-gap constructions in spoken English corpora -- matrix wh-questions, embedded wh-questions, and relative clauses -- and further identifies the extraction site (i.e., subject vs. object vs. adjunct). Our approach combines constituency and dependency parsing, leveraging their complementary strengths for construction classification and extraction site identification. We validate the system on human-annotated data and find that it scores well across most categories. Applying the system to 57 English CHILDES corpora, we are able to characterize children's filler-gap input and their filler-gap production trajectories over the course of development, including construction-specific frequencies and extraction-site asymmetries. The resulting fine-grained labels enable future work in both acquisition and computational studies, which we demonstrate with a case study using filtered corpus training with language models.