The bulk of our knowledge of language change comes from Indo-European languages, for which we have a vast historical record, and a long tradition of studies. Minority, typologically diverse languages — such as indigenous American ones — are key to a broader understanding of change in language as a whole, yet, where relevant material is available, it tends to be woefully under-studied. In this context, nding ways to systematically compile and explore the available data is crucial. In this talk I will focus on the Native American language Mapudungun (presumed-isolate, Chile/ Argentina), for which we have written records dating back to the early seventeenth century (Valdivia 1606, 1621, Herckmans 1642-1643). While there is very little explicit work on the lan- guage’s history, current varieties are fairly well described. Mapudungun is particularly interesting to our overall understanding of language change insofar as it is typologically very distinct from IE languages. Features such as incorporation, root serialisation, agglutination and polysynthesis raise interesting questions about the diachrony of units of sound and meaning, which cannot be probed by IE languages. Here I will give some background on Mapudungun and its history, to then move on to some of the challenges — for data, methods and theory — in the study of its development. I will introduce my current project to build a Corpus of Historical Mapudungun using a substantial selection of Mapudungun texts from the 17th to the 20th century, and exemplify its use by exploring the diachrony of vowel epenthesis and stress. I conclude with some remarks on whether such changes — in the context of asymmetric language contact — should be viewed as endogenous, or contact- induced.