I study artificial neural network models for natural language understanding, with a focus on language model alignment, building high-quality training and evaluation data, and applying neural network models to scientific questions in syntax and semantics.