How poor is the stimulus? Evaluating hierarchical generalization in neural networks trained on child-directed speech
Preprint
Abstract
When acquiring syntax, children consistently choose hierarchical rules over competing non-hierarchical possibilities. Is this preference due to a learning bias for hierarchical structure, or due to more general biases that interact with hierarchical cues in children's linguistic input? We explore these possibilities by training LSTMs and Transformers - two types of neural networks without a hierarchical bias - on data similar in quantity and content to children's linguistic input: text from the CHILDES corpus. We then evaluate what these models have learned about English yes/no questions, a phenomenon for which hierarchical structure is crucial. We find that, though they perform well at capturing the surface statistics of child-directed speech (as measured by perplexity), both model types generalize in a way more consistent with an incorrect linear rule than the correct hierarchical rule. These results suggest that human-like generalization from text alone requires stronger biases than the general sequence-processing biases of standard neural network architectures.Keywords
All Related Versions
This publication has 29 references indexed in Scilit:
- Using Computational Models to Test Syntactic LearnabilityLinguistic Inquiry, 2024
- Poverty of the Stimulus Without TearsLanguage Learning and Development, 2021
- Meaningful questions: The acquisition of auxiliary inversion in a connectionist model of sentence productionCognition, 2017
- Syntactic Islands and Learning Biases: Combining Experimental Syntax and Computational Modeling to Investigate the Language Acquisition ProblemLanguage Acquisition, 2013
- Poverty of the Stimulus RevisitedCognitive Science, 2011
- Linguistic Nativism and the Poverty of the Stimulus2011
- Bigrams and the Richness of the StimulusCognitive Science, 2008
- Expectation-based syntactic comprehensionCognition, 2008
- Is Structure Dependence an Innate Constraint? New Experimental Evidence From Children's Complex‐Question ProductionCognitive Science, 2008
- Identification in the Limit of Substitutable Context-Free LanguagesLecture Notes in Computer Science, 2005