This TICS article by Pinker and Ullman, with replies by McClelland and colleagues, may be of interest as a place to dive into the debate that the articles we are now reading started. Also, this reply by Embick and Marantz to another article by Ullman and colleagues highlights the fact that the lexicon/rule distinction in Pinker and Ullman is not assumed in quite this fashion in generative theories of morphophonology. And following up on Alex’s comment, here’s the empirical study on child overregularization I believe he was referring to, and a connectionist modeling reply.