Text Mining Oral Histories in Historical Archaeology

Abstract

Advances in text mining and natural language processing methodologies have the potential to productively inform historical archaeology and oral history research. However, text mining methods are largely developed in the context of contemporary big data and publicly available texts, limiting the applicability of these tools in the context of historical and archaeological interpretation. Given the ability of text analysis to efficiently process and analyze large volumes of data, the potential for such tools to meaningfully inform historical archaeological research is significant, particularly for working with digitized data repositories or lengthy texts. Using oral histories recorded about a half-century ago from the anthracite coal mining region of Pennsylvania, USA, we discuss recent methodological developments in text analysis methodologies. We suggest future pathways to bridge the gap between generalized text mining methods and the particular needs of working with historical and place-based texts.

Anthracite memories: semantic tagging and coal mining oral histories

“…Our custom stopword list included terms such as 'interviewed', 'they'd', 'yeah', and the names of the interviewers and interviewees. The full stopword list is included in the Appendix and article GitHub repository (Brown and Shackel 2023). Finally, we further standardised the words and bigrams in terms of tense, parts of speech, and plurality.…”

“…Annotating these terms based on expertise in the field and then using these categories as a framework for further analyses combines the qualitative and quantitative aspects of interpreting texts. In support of open and reproducible science, these tagging libraries and stopword lists are available in the Appendix and online in the project GitHub repository (Brown and Shackel 2023).…”