Multivariate modeling of the collaboration between Luigi Illica and Giuseppe Giacosa for the librettos of three operas by Giacomo Puccini

摘要

This article explores the collaboration between the two Italian poets Luigi Illica and Giuseppe Giacosa that led to the librettos of three operas by Giacomo Puccini: La Boheme, Tosca, and Madama Butterfly. We used non-traditional authorship attribution methods (such as principal component analysis) and pattern recognition methods (such as partial least-squares discrimination analysis) to investigate whether traces of the two authors' personal writing styles could be highlighted in the final output of their collaboration. Working in a closed authorship attribution problem framework, we show that the influence of Luigi Illica on the versification of these librettos and their general poetic tone may be even greater than suggested by recent traditional scholarly studies. Vocabulary analysis carried in a natural language processing framework suggests that Illica contributed not only (and perhaps not mainly) to the drama but also significantly to the poetic tone of the librettos. Novel directions in Puccini scholarly research emerge from the results of our computational approach to the Illica-Giacosa collaboration.