Inductive Linguistic Reasoning with Large Language Models