Copyright Is the Lock; Non-Expressive Fair Use Is the Key: Research with In-Copyright Texts

Copyright restrictions have long been understood as persistent and seemingly insurmountable barriers for digital scholarship in those many research areas that focus on texts from the copyright era. But a series of recent developments in the digital library world, in digital humanities research, and in the legal sphere have led to innovative legal and practical concepts, among which "non-expressive" (otherwise known as "non-consumptive") use is perhaps the most visible and productive. This concept, along with technical and infrastructural practices supporting it, have enabled the massive opening of library materials for digital research - including those that had previously been restricted by copyright concerns. This chapter describes the history, principles, and uses of one extremely important example of such an embodiment of non-expressive research practice in digital libraries: the HathiTrust Research Center's Extracted Features Dataset, a staple of text data mining work in DH.