Wednesday, November 20, 2019

Analyzing Co-Authorship in Timon of Athens using the Shakespeare Affinity Test

It has been suggested that Shakespeare's play Timon of Athens might have been co-authored with Thomas Middleton. Here I will use some techniques made possible by my Shakespeare Affinity Test to investigate this question.

Plotting the Unusual Shakespeare Words

As I explained in my blog post, the Shakespeare Affinity Test identifies words that occur in Shakespeare plays but are relatively unusual compared with the rest of the database of 500+ plays. By charting where these words occur in a play possibly co-authored by Shakespeare, it might be possible to determine which sections were written by Shakespeare and which sections were written by someone else.

Note, the Shakespeare Affinity Test is not an authorship test. There is currently no reliable method for determining authorship. The SAT is just a tool for analyzing the unusual vocabulary in a play.

I ran Timon of Athens using standard parameters of the SAT to create this chart that plots these "hits":


As you can see above, there are several gaps where there are fewer rare Shakespeare words. These are:

Gap 1: Words 797-1799. Roughly Act 1, Scene 1, Line 119-275.
Gap 2: Words 8945-10325. Essentially all of Act 3, Scene 5 and 6.
Gap 3: Words 14754-16259. Most of Act 5, Scene 1.

Comparing with Wikipedia's summary of one person's analysis:
John Jowett, editor of the play for both the Oxford Shakespeare: Complete Works and the individual Oxford Shakespeare edition, believes Middleton worked with Shakespeare in an understudy capacity and wrote scenes 2 (1.2 in editions which divide the play into acts), 5 (3.1), 6 (3.2), 7 (3.3), 8 (3.4), 9 (3.5), 10 (3.6) and the last eighty lines of 14 (4.3).
The only agreement between the above chart and this analysis seems to be on Act 3, Scene 5 and 6.

Analyzing Unusual Words in the Gaps

I ran a rare word test on just the gaps. These are rare words that occur in the gaps, no other Shakespeare play, and only 20 or fewer times in the whole database:

unclew, 1354 -- unique
repugnancy, 9067 -- The Duchess of Suffolk
byzantium, 9181 -- Selimus, Hans Beer-Pot
briber, 9185 -- unique
rioter, 9235 -- Mucedorus; Middleton, A Trick to Catch the Old One (twice) and Michaelmas Term; Yorkshire Tragedy
usure, 9579 -- common word
exceptless, 15162 -- unique
usure, 15260 -- common word
phrynia, 15531 -- unique
timandra, 15533 -- The Bondman, Philip Massinger (1624)
opulency, 15798 -- unique

If Middleton wrote these sections, you might expect there to be words not found in other Shakespeare plays that Middleton commonly used. However, there is little evidence here for Middleton's authorship of these passages. The only word of interest is "rioter" in Act 3, Scene 5 of the play.

Analyzing Unique Words in Timon of Athens

Using the database, we can easily generate a list of all of the words that appear in Timon of Athens but in no other Shakespeare First Folio plays. There are 91 such words:

alcibiades, apemantus, approacher, ardent, balsam, banditti, blain, brevis, briber, byzantium, caphis, carper, castigate, cauterize, composture,, confectionary, contentless, decimation, defiler, detention, distasteful, dividant, droplet, embalm, ensear, exceptless, exhaust, flaminius, fragile, furor, gluttonous, hortensius, indisposition, insculpture, ira, isidore, jutting, lacedaemon, liquorish, lucullus, madwoman, mangy, manslaughter, misanthropos, monstrousness, mountant, numberless, nutriment, oathable, obliquy, opulency, passive, pelf, penurious, philotus, phrynia, procreation, rampire, recanter, recoverable, regardful, regular, reliance, repugnancy, rioter, sacrificial, servilius, slavelike, softness, solidar, spilth, steepy, straggle, suitable, timandra, towardly, trenchant, unagreeable, unaptness, uncharged, unclew, unctuous, unpeaceable, untirable, usure, varro's, viced, wappened, whittle

We can use this as a basis to compare with the other plays in the database. I ran a test on five of Middleton's plays from the period; of those 91 words, these are the ones that showed up in those plays:

The Phoenix (1603-4) -  balsam
Michaelmas Term (1604) - rioter, towardly
Trick to Catch the Old One (1605) - penurious, rioter, usure
Mad World - none
A Chaste Maid in Cheapside (1613) - 37 - suitable

"penurious" was pretty common in plays of the period as was "balsam". "towardly" was in Eastward Ho, so Middleton's use isn't notable. "usure" was relatively common; it's in Volpone and other plays.

Really the only word of interest is "rioter". As explained above, that word appears in Act 3, Scene 5 of Timon of Athens, in one of the gaps and was not a very common word in plays of the period. Middleton uses it in two plays of the period and if he actually wrote Yorkshire Tragedy, in three plays. 

Comparing Relatively Rare Words in Middleton's plays with Timon of Athens

Now I will do a mini "Middleton Affinity Test". I will select out all of the unusual words in the above five plays and see how often and where they occur in Timon of Athens. Some of these words occur in other Shakespeare plays. Here is the list:

6185 scarcity
6538 towardly
7121 disfurnish
8885 unnoted
9235 rioter
9579 usure
11539 spital
12502 quillet
14682 thievery
14691 attraction
15260 usure

Most of these words we looked at above and the other ones are not particularly notable. "spital" is common in Shakespeare; "attraction" is in Merry Wives of Windsor and Pericles; "scarcity" is in Venus and Adonis; "quillet" is very common; "unnoted" is in Rape of Lucrece. "disfurnish" is in Two Gentlemen of Verona and Pericles.

This list actually shows the likelihood that Middleton borrowed some unusual words from Shakespeare rather than anything else. It certainly is not evidence of co-authorship.

Conclusion

There is no reason to conclude from this that Thomas Middleton wrote a substantial portion of Timon of Athens. There are some gaps in rare Shakespeare words, particularly in Act 3, Scene 5 and 6. If you want to look for co-authorship, I would look there.  The attribution to Middleton is possible, but seems largely unsupported by this data.