MICASE is a sizeable (“nearly 1.8 million words”) corpus of spoken, academic English. It may not be the biggest corpus around, and the website doesn’t offer fancy tools like COCA does, but what’s great about it is that you can be SO SPECIFIC about the speaker and context. When you run a search, you can choose the gender, age, academic position/role, native speaker status and first language of the speakers, and also the context: Speech Event Type, Academic Division, Academic Discipline, and “Interactivity Rating” (e.g. monologic vs. highly interactive.” Also, for some of it, you can even LISTEN to the data. Thanks, alma mater!