Computational Stylometry of Wittgenstein’s “Diktat für Schlick”.

Resumen

Using techniques from computational stylometry we will examine some of the dictated writings of Ludwig Wittgenstein which have been made available by the Wittgenstein Archives at the University of Bergen. Our purpose is to give an example of how computational stylometry can be used to help answer concrete questions of Wittgenstein research, and thus to explore how computational stylometry can be applied to issues of philosophical authorship more generally.  In particular we use computational stylometry to help examine the question of whether the so-called "Diktat für Schlick" was in fact dictated by Wittgenstein, and if yes, whether to Schlick or Waismann. To this end we compare documents known to be written by Wittgenstein in person; the "Diktat für Schlick"; and samples of some of Schlick's and Waismann's (the potential amanuenses) own writings. Using the "Stylometry with R" package of Eder and Rybicki, the degree of similarity and dissimilarity between documents is calculated by Burrows' delta measure, and the results are displayed using Cluster Analysis, Multidimensional Scaling and Bootstrap Consensus Trees.  The documents are each characterised by the frequencies of the 300 most frequent words in the entire corpus, normalised by document length. Our main findings are that the "Diktat für Schlick" is far more similar to Wittgenstein's writings than those of either Schlick or Waismann, and that the writings of Wittgenstein and Schlick or Waismann are stylometrically quite distinct.