Minimally supervised model of early language acquisition

Abstract

Theories of human language acquisition assume that learning to understand sentences is a partially-supervised task (at best). Instead of using 'gold-standard' feedback, we train a simplified "Baby" Semantic Role Labeling system by combining world knowledge and simple grammatical constraints to form a potentially noisy training signal. This combination of knowledge sources is vital for learning; a training signal derived from a single component leads the learner astray. When this largely unsupervised training approach is applied to a corpus of child directed speech, the BabySRL learns shallow structural cues that allow it to mimic striking behaviors found in experiments with children and begin to correctly identify agents in a sentence. © 2009 Association for Computational Linguistics.