Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Zhenghao Herbert Zhou, William Dai, Maya Viswanathan, Simon Charlow, R. Thomas McCoy, Robert Frank

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.

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

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.
Paper Structure (25 sections, 13 figures, 6 tables)

This paper contains 25 sections, 13 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Target Constructions
  3. Detection Algorithms
  4. Detection Steps
  5. Step 1: Core Structure Detection
  6. Step 2: Wh-category Identification
  7. Step 3: Extraction Site Inference
  8. Step 4: Dependency Validation
  9. Arguments for a hybrid approach
  10. Evaluation
  11. Manual Evaluation
  12. Comparison with pearl2013computational
  13. A Case Study on CHILDES
  14. Distributions Across Time
  15. Total Exposures of One Individual Child
  16. ...and 10 more sections

Figures (13)

  • Figure 1: Dependency (top) and constituency (bottom) parses of a sentence containing a subject relative clause.
  • Figure 2: Dependency parses of an embedded question (left) and a matrix question (right).
  • Figure 3: Constituency parses of an embedded question (left) and a matrix question (right).
  • Figure 4: Adult (solid) and child (dashed) speech distributions across time by constructions and extraction sites. Uncertainty is shown as 95% Wilson intervals (wider in sparser bins).
  • Figure 5: Per-1,000 utterance rates of all filler-gap dependency sentences received and produced by Laura.
  • ...and 8 more figures