Reflections on reproducible research

Abstract

Reproducibility in language documentation and description means that the analysis given in descriptive publication is presented in a way that allows the reader to access the data on which the claims are based, to verify the analysis for themself. Linguists, including Himmelmann, have long pointed to the centrality of documentation data to linguistic description. Over the twenty years since Himmelmann’s 1998 paper we have seen a growth in digital archiving, and the rise of the Open Access movement. Although there is good infrastructure in place to make reproducible research possible, few descriptive publications clearly link to underlying data, and very little documentation data is publicly accessible. We discuss some of the institutional roadblocks to reproducibility, including a lack of support for the development of published primary data. We also look at what work on language documentation and description can learn from the recent replication crisis in psychology.