Home
Philosophy and Text Technology
Foredrag

Identifying and Interpreting Similarities in Wittgenstein's Nachlass

To what extent can (or cannot) computer-based methods for discovering similarities in the Nachlass replicate or supplement the results of the traditional approach of closely reading, interpreting and comparing texts?

Main content

Wittgenstein's Nachlass includes a large number of revisions, resulting in often only slightly different versions of the same text. Readers looking primarily for Wittgenstein's latest or most considered views on a particular subject may be frustrated. Readers interested in the development of Wittgenstein's thought may be fascinated. Both kinds of readers may find information about similarities in the Nachlass useful. And indeed, such relationships between various parts of the Nachlass have been studied extensively. 

This talk will report on a study of some computer-based methods for discovering similarities in the Nachlass. To what extent can (or cannot) such more formal methods replicate or supplement the results of the traditional approach of closely reading, interpreting and comparing texts? The methods employed are so-called bag-of-word analyses of n-grams in combination with lemmatization and a number of different similarity measures set to different score thresholds. Pairs, clusters and cliques of similar pieces of text are identified and compared. Earlier work by Schulte, Maury and WAB are used to calibrate the choice of parameters, and results are compared also to other text corpora.