Revisiting the poverty of the stimulus: hierarchical generalization without a hierarchical bias in recurrent neural networks
This addresses the poverty of the stimulus problem in language acquisition by showing that implicit architectural biases and input cues may suffice for hierarchical learning, offering insights for computational linguistics and cognitive science.
The study investigated whether recurrent neural networks (RNNs) can learn hierarchical syntactic rules for question formation in English without explicit hierarchical biases, finding that some RNN architectures successfully acquired these rules, with success rates increasing when subject-verb agreement cues were included.
Syntactic rules in natural language typically need to make reference to hierarchical sentence structure. However, the simple examples that language learners receive are often equally compatible with linear rules. Children consistently ignore these linear explanations and settle instead on the correct hierarchical one. This fact has motivated the proposal that the learner's hypothesis space is constrained to include only hierarchical rules. We examine this proposal using recurrent neural networks (RNNs), which are not constrained in such a way. We simulate the acquisition of question formation, a hierarchical transformation, in a fragment of English. We find that some RNN architectures tend to learn the hierarchical rule, suggesting that hierarchical cues within the language, combined with the implicit architectural biases inherent in certain RNNs, may be sufficient to induce hierarchical generalizations. The likelihood of acquiring the hierarchical generalization increased when the language included an additional cue to hierarchy in the form of subject-verb agreement, underscoring the role of cues to hierarchy in the learner's input.