Can function word distribution indicate authorship?
James Larkin Wyatt · ACM SIGLASH Newsletter · 1980
The writer's interest in the answer to the question contained in the title of this paper stems from his discovery of seemingly startling aberrations in word frequency distributions in the Celestina, an anonymous work of Spanish literature which conceivably could be the product of one, two, or three authors. This work appeared in 1499 with sixteen parts. An edition in 1501 also appeared with sixteen parts but contained a letter "to a friend" stating that the first part was written by one unnamed person and that the succeeding fifteen parts were written by the author of the letter, identified acrostically as Fernando de Rojas in some verses forming a part of the letter. An edition in 1502 contained five additional parts inserted between the fourteenth and fifteenth original parts, and the original fourteenth part was considerably amplified. There has been much discussion and debate, but the consensus today is that there was a single author of the work.