This paper discusses the digital dictionary component in an ongoing language documentation project for the Mixtepec-Mixtec language (iso 639-3: mix). Mixtepec-Mixtec (Sa'an Savi 'rain language') is an Oto-monguean language spoken by roughly 9,000-10,000 people in the Juxtlahuaca district of Oaxaca Mexico. Creating a digital dictionary for an under-resourced language entails a number of challenges that require unique and nuanced encoding solutions in which a delicate balance between the linguistic content, data structure, potential linked resources, and editorial metadata must be found. Herein we demonstrate how we use TEI to create a reusable, extensible, and machine readable language resource with an emphasis on how our solutions using a combination of novel and established TEI dictionary structures enable us to address our specific needs for Mixtepec-Mixtec and also provide a relevant roadmap for similar under-resourced language projects.