In Shakespeare’s time, about a quarter of all plays were collaboratively written by two or more dramatists. Christopher Marlowe’s classic work Doctor Faustus was first performed in the 1580s or early 1590s but only published in 1604, 11 years after his death. The dramatists Samuel Rowley and William Bird were paid in 1602 to write new additions to the play.
However, researchers have long suspected that the original play was not written by Marlowe alone. Its broader comic parts, largely in prose, have been thought to be the work of a second writer who never got the credit they deserved.
Previous research has suggested one of the dramatists Thomas Nashe or Henry Porter. Nashe left us only one play, Summer’s Last Will and Testament (1592). Porter wrote several in collaboration with other dramatists, but only one sole-authored play survives, The Two Angry Women of Abingdon (1599).
So, can a new computer analysis of texts resolve this question and finally credit the second author of Doctor Faustus?
In an article in the journal Notes & Queries and a piece in the Guardian, lecturer in Shakespeare studies Darren Freebury-Jones describes computer techniques that point to Marlowe’s co-author being Porter. But we have found methodological problems in Freebury-Jones’s approach, and believe the evidence he presents does not support the conclusions he draws from it.
Computational analysis can certainly generate new knowledge about authorship, but his work is not a convincing example of this burgeoning art.
Unique matches
Freebury-Jones’s Notes & Queries article presents tables from a free online dataset, Collocations and N-Grams, created by independent Shakespeare analyst Pervez Rizvi. For 527 plays from the 1550s to the 1650s (pretty much all of those that have survived from this period), the dataset lists all four-word phrases that each play has in common with the other 526.
But the phrases Rizvi considers most significant are those that appear only in the play under consideration and one other play – known as “unique matches”. For each of the 527 plays, he ranks which other plays have the most unique matches to it.
If these rare phrases are particular to each dramatist’s idiolect (unique way of speaking), then researchers should, for example, find at the top of the Macbeth table some other plays by Shakespeare. And we do: among the top-ten plays that share the most unique matches with Macbeth are Julius Caesar, Coriolanus, King Lear and Hamlet.
But while the top-ten list for Macbeth gives grounds for optimism about Freebury-Jones’s method, since four of its plays are Shakespeare’s, the top-ten list for Doctor Faustus, reproduced below, dashes this:
- The Taming of a Shrew
- A Trick to Catch the Old One
- The Devil’s Charter
- Albumazar
- Caesar and Pompey, or Caesar’s Revenge
- The Family of Love
- Messalina, the Roman Empress
- The Sisters
- The Two Angry Women of Abingdon
- Tamburlaine
While Porter is present in this list as the author of The Two Angry Women of Abingdon, Marlowe himself has only one play in it: Tamburlaine (1587) at position ten. If this method were detecting authorship, Marlowe’s other plays should dominate the list, just as the Macbeth list is dominated by Shakespeare plays.
Moreover, Porter’s play also appears at number ten for unique matches with Macbeth. This suggests the evidence for Porter being a coauthor of Macbeth is almost as strong as for Doctor Faustus – but Freebury-Jones makes no mention of this.
Porter can hardly be expected to dominate any top-ten list as Shakespeare does for Macbeth, since he left us only play. In contrast, with his roughly 1 million words of writing, Shakespeare presents a much greater “surface area” to any method that counts rare words or phrases – so he has more opportunities than anyone else to write the rare phrases we are looking for. To compensate for this, Rizvi discounts the significance of matches to authors with large canons.
Freebury-Jones mentions that the raw counts “are divided by the combined word counts for each pairing”, which is the scaling process Rizvi describes in the notes to Collocations and N-Grams. But nobody knows if this is the right way to scale for different canons or not, because the science of this problem has not been explored.
Rizvi changed his scaling formula in 2018 at my (Gabriel Egan’s) suggestion. But I did not claim to have solved the problem, and I consider it an open question whether we can derive a fair formula for such scaling.
Furthermore, Rizvi’s dataset does not distinguish between the plays that Shakespeare wrote alone and those he co-wrote, so his scaling factor treats as Shakespeare’s some substantial writings of other authors.
Further problems
This highlights another question with the analysis. In the top-ten list for unique matches with Macbeth are two plays co-authored by the dramatists Francis Beaumont and John Fletcher: The Woman Hater (1606) and Philaster (1620).
So, what should a researcher do when a play has more than one author? One response is to count matches to Beaumont’s part separately from matches to Fletcher’s part, using the widely accepted divisions of these two plays – as first proposed by Cyrus Hoy, an expert in the English Renaissance stage, in 1958. Alternatively, you could exclude co-authored plays entirely, as many authorship investigators do.
But Freebury-Jones says nothing about this problem of judging writers by their co-authored works. This consideration bears somewhat on Macbeth too, since many Shakespearians believe the playwright Thomas Middleton wrote about 5%-10% of it.
So where does all this leave Porter? Freebury-Jones ends his Notes & Queries piece with a cautious claim: “We may reasonably conclude that Henry Porter is a likelier candidate than [Thomas] Nashe for Marlowe’s collaborator on Doctor Faustus.”
But in speaking to the Guardian, he abandoned this caution. He reports being “astonished by just how close the dramatic language of Doctor Faustus actually was to Porter”, and insists that we must recognise “Porter as the most likely collaborator”.
It’s possible that Porter did co-write Doctor Faustus. But the problems with Freebury-Jones’s analysis mean that, for now, we still can’t say for sure.
Looking for something good? Cut through the noise with a carefully curated selection of the latest releases, live events and exhibitions, straight to your inbox every fortnight, on Fridays. Sign up here.
The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.
This article was originally published on The Conversation. Read the original article.
