You are invited to the next Philosophy Seminar Series on Tuesday 3 October, 3-5pm – Beatrice Tinsley 329, University of Canterbury
‘Philosophy in Historical Newspaper Archives: Content-First vs. Practice-First Labelling’ by Joshua Black, University of Canterbury
Abstract: Historical newspaper archives represent a significant source of data for experimental philosophers and historians of philosophy. However, the promise of such archives comes with a series of serious practical hurdles. The ‘needle in the haystack’ problem is a major problem for researchers who wish to explore philosophical discourse in historical newspapers.
Newspapers are, unsurprisingly, dominated by news reporting, advertising, public notices and so on. This is especially problematic for those who wish to apply more sophisticated text analytic methods, which will require a great deal of computational resources to pick up any relevant signal in such large archives. In previous work, I present a method for generating specialised corpora, using a study of philosophical discourse in 19th Century English-Language New Zealand newspapers (Wilson Black, 2022).
This method uses a ‘content-first’ labelling method. It begins with my own judgements as to what constitutes philosophical content. Once labels have been assigned, a text classifier is then trained to pick out philosophical items from the complete archive. In this paper, I introduce an alternative ‘practice-first’ approach, which labels advertisements of public lectures, uses a ‘fuzzy search’ method to pick out such reports, and then applies an information extraction model to obtain a bottom-up sense of the content covered in public lectures. This approach picks out philosophy in large newspaper archives by starting with a philosophically relevant practice: the giving. The advantages and disadvantages of each method are considered with respect to the kinds of research questions which an historian of philosophy might have.
Zoom link: (passcode: 1).
