The Importance of Category Labels in Grammar Induction with Child-directed Utterances

TL;DR

This paper demonstrates that incorporating category labels in grammar induction evaluation reveals important linguistic insights and confirms the effectiveness of depth-bounding in modeling human memory constraints across multilingual child-directed speech.

Contribution

It introduces labeled evaluation metrics for grammar induction and shows that depth-bounding remains effective with labeled data in multilingual contexts.

Findings

Labeled evaluation uncovers linguistically relevant grammar properties.

Depth-bounding improves grammar induction on child-directed speech.

Labeled metrics reveal limitations of unlabeled evaluation methods.

Abstract

Recent progress in grammar induction has shown that grammar induction is possible without explicit assumptions of language-specific knowledge. However, evaluation of induced grammars usually has ignored phrasal labels, an essential part of a grammar. Experiments in this work using a labeled evaluation metric, RH, show that linguistically motivated predictions about grammar sparsity and use of categories can only be revealed through labeled evaluation. Furthermore, depth-bounding as an implementation of human memory constraints in grammar inducers is still effective with labeled evaluation on multilingual transcribed child-directed utterances.