Summary: | This article describes a potential new research resource for looking at the development of archaeological practice longitudinally, from its roots in antiquarianism through to the modern discipline. As a result of a Natural Language Processing (NLP) exercise carried out at the Archaeology Data Service (ADS) we have a rich dataset comprising frequency counts of place and period names from a corpus of historical archaeological journals. This potentially provides a tool to examine the reflexive nature of the relationship between archaeology and broader society, particularly with regard to what aspects of the past, periods and areas have prominence at various times in the development of the discipline. For example, while it is well recognised that there are strong links between interest in the Roman Empire and the construction of an imperial vision of Britain in the Victorian age, demonstrating this metrically is a laborious and time-consuming activity. Similarly, conclusively identifying bias towards a particular time period or geographic region can be difficult to do with a hundred and fifty years' worth of journals to work with. This can mean that the sometimes subtle, transient notions of what, when and where has been significant in our past can be hard to critique retrospectively.
The analysis of what political or social processes lead to a concentration of interest and resources on particular monuments, areas or time periods is similarly problematic. Often the archaeological work that results from these processes becomes the source of a cycle whereby these time periods and places appear to become more important than others simply because they have been investigated and described. This material itself thereby provides the fuel for ongoing dialogue, additional investigation and further publication. At the same time, other areas and time periods assume a secondary importance and come to occupy a peripheral place in the broader narrative of the past.
These new data comprise metric data garnered through 'data mining' and NLP techniques on a substantial corpus of legacy literature. The availability of these data is actually a by-product of another process designed to allow the indexing and easy discovery of archaeological text by researchers under the auspices of the 'Archaeotools' project. This project was a major e-Science project, and a joint endeavour between the ADS based at the University of York and the Natural Language Processing Research Group at the University of Sheffield funded by the UK's Arts and Humanities e-Science Initiative e-Science Research Grants Scheme, which itself was a collaboration between three major funding bodies, the Arts and Humanities Research Council (AHRC), the Engineering and Physical Sciences Research Council (EPSRC) and the JISC. As well as describing this new resource, this short article presents some possible broad conclusions that exemplify the potential of the resource and suggests directions for further analysis.
|