SHAKSPER 2005: Lions and Tigers and Wagers...oh my...

From: Hardy M. Cook (editor@shaksper.net)
Date: 11/27/05 

The Shakespeare Conference: SHK 16.1951  Sunday, 27 November 2005

From: 		Ward Elliott <WElliott@claremontmckenna.edu>
Date: 		Saturday, 26 Nov 2005 13:40:55 -0800
Subject: 16.1933 Lions and Tigers and Wagers...oh my...
Comment: 	RE: SHK 16.1933 Lions and Tigers and Wagers...oh my...

It seems that the dust hasn't settled on the Woodstock controversy after 
all.  William Davis (SHK 16.933, Nov. 23) wonders whether our 
stylometric data, though well enough validated for Shakespeare's "core" 
plays written "mostly at the height of his career," can tell us anything 
worthwhile about whether a play written "very early in his career," or 
"pre-career," could be identified as his.  Mr. Davis doesn't specify 
what dates he means by "very early" or "pre-career," but his point of 
reference is our "near-wager" with Michael Egan, who last summer claimed 
to have proved irrefutably that Woodstock must have been written by 
Shakespeare by 1594-95, challenged us to a showdown on the question, but 
in the end declined our offer to put it to a vote by SHAKSPER's 
membership.

The Woodstock reference is important for two reasons: (1) since there is 
no external evidence connecting the play to Shakespeare, the debate is 
entirely over internal evidence; and (2) Mr. Egan, like most scholars 
prior to 2002, strongly supports a date between 1592 and 1595, just the 
right time for a Richard II, Part I to have been written.  We can only 
speculate whether and how much a hypothetical "pre-career" play or poem 
written by Shakespeare in his teens or early 20's might resemble works 
he wrote in his 30's and 40's.  We discuss this point in detail in our 
"Oxford by the Numbers" (our 2004, pp. 382-396) with respect to the 
"pre-career" poems of the Earl of Oxford and conclude that it is 
implausible for many reasons that a stylometric sow's ear like Oxford 
would mature into a silk purse like Shakespeare overnight in his mid-40's.

But Woodstock is hardly a "pre-career" play; in fact, it's probably not 
even a "very early" play.  MacDonald Jackson notes that it's a 
17th-century scribal manuscript loaded with 17th-century words in 
17th-century meter and concludes that it probably was written in the 
17th century when Shakespeare was in his late 30's or early 40's, and 
not by Shakespeare, but more likely by Samuel Rowley (Jackson, 2002). 
We think Jackson's arguments are persuasive but for now are willing to 
accept the older, conventional, Egan dating for argument's sake.  This 
could make it, arguably, "very early" for a Shakespeare play, written 
just after Richard III (1592-93) and the Henry VI series (1589-91) when 
Shakespeare was nearing 30.  So let's take Woodstock, Richard III, and 
Henry VI, Part II, thought to be the least co-authored of the H6 series 
and the one by which Woodstock was chiefly inspired. Let us suppose that 
no one had heard of any of these, and that all three were discovered as 
anonymous plays in Mr. Davis's time capsule.  Is there any way you could 
tell from Shakespeare's remaining core plays which of the three 
manuscripts was Shakespeare's?

We would expect Mr. Egan to say the three plays were indistinguishable 
from each other, under some variant of his past arguments from 
resemblances that Woodstock, at least, has to be Shakespeare's.  He 
thinks it dovetails so perfectly with Richard II that it has to be the 
first part of a sequence, Richard II, Part I, and that its 1,600 
"unique" Shakespeare "fingerprints" prove irrefutably that Shakespeare 
wrote it.  Shakespeare spoke of "heads cut off," of "forfeiture of 
land," and of "nearness in blood;" so did whoever wrote Woodstock; 
therefore Shakespeare must have written Woodstock.   We don't find such 
parallels persuasive because we doubt that they are unique, and we 
actually searched a couple of them - but not all 1,600 -- in 
contemplation of the later-rejected debate, and we found many parallels 
in other writers.  To us, they are more like fingers, toes, eyes, and 
ears than like fingerprints or DNA.  If you and Shakespeare both have 
ten fingers and two ears, we don't think it necessarily proves that you 
are Shakespeare.

Our preferred approach is to look for differences from Shakespeare, not 
resemblances, because differences disprove common authorship a hundred - 
or is it a thousand? - times more powerfully than non-unique 
resemblances prove it.  If you and Cinderella both fit a Size Four 
slipper, it doesn't prove you are she, only that you could be.  But if 
you are a Size Eight, your claim is in real trouble.  If there are only 
a few of such differences in a sample text, it could be Shakespeare's; 
if many, it is that much less likely.

What would happen to our core Shakespeare baseline profiles if we had 
never heard of either R3 or 2H6?  The short answer is "nothing." 2H6 is 
not in our core baseline because we wanted a clean baseline and were 
aware that most scholars suspect that the entire H6 series is 
co-authored.  R3 is a gold-standard baseline Shakespeare play, but not 
one that sets any boundaries.  On every test it scores within profile 
boundaries set by the other 28 core Shakespeare plays.  For our 
time-capsule thought experiment that means that we don't have to change 
a single profile to test the three "anonymous" plays.

The first and least important outcome of such a test is this:  all three 
plays, like all our Shakespeare plays and most of our plays not by 
Shakespeare, would pass at least 28 of the 48 tests.  They all have 
plenty of Shakespeare resemblances.  But these piles of non-unique 
resemblances prove very little.  Much more important is the second 
outcome, marked by stars (*) in Table 1 below: Woodstock fails all of 
the remaining 20 tests, while R3 fails none and 2H6 fails only three, 
with a Shakespeare composite threshold of two.  The odds that 
Shakespeare could have written Woodstock are trillions of times lower 
than the odds that he could have written R3 or 2H6, no matter which of 
our two odds-calculating conventions is used.  See our 2004, pp. 399, 
402.  In terms of composite distance from Shakespeare's normal ranges, 
our "very early" 2H6 is just outside our Shakespeare ballpark, and our 
"very early" R3 is in the infield just a yard or two from home plate- 
but the "very early" Woodstock is in a different statistical galaxy. 
These seem to us very powerful prima facie evidence that R3 certainly, 
and 2H6 possibly, are Shakespeare could-be's, but Woodstock is not.

Table 1
Shakespeare Rejections, Richard III, 2H6, and Woodstock

Shakespeare Test, Range		R3	2H6	Woodstock	Remarks
Grade Level, 4 to 7		5	6	3*		g, e
Hyphenated Words, 52 to 180	85	82	8*		e
Feminine Endings, 8 to 17	16	12	18*		t, p
Open Lines, 11 to 23		17	14	32*		t, p, e
It(1st wd of sentence),7-18	9	9	4*		e
BoB1, 284 to 758			425	284	201*	
BoB3, -174 to 247			-109	-179*	-202*	
BoB7, 278 to 779			358	514	32*	
Aggregate Buckets, -2 to +2	-0.95	-1.54	-5.1*	
Total 1, 0 to 9			0	1	28*	
On't per 20,000, 0 to 2		1	0	6*		t
i'faith per 20,000, 0 to 8	0	1	9*	
Ha' per 20,000, 0 to 5		0	0	22*	
'll per 20,000, 31 to 90	41	55	136*	
'd/'ld per 20,000, 0 to 2	0	1	11*		t
I do + verb, 5 to 28		13	11	2*	
-ish per 20,000, 1 to 22	12	8	23*	
-ly per 20,000, 98 to 161	155	108	194*	
Most + modifier, 8 to 32	18	17	34*		t
See per  20,000, 0 to 5		4	3	7*	
Whereas/whenas per 20,000,0	0	2*	0		e
It,last wd/s per 20,000, 8-30	16	7*	17	
				
Midl. sp. end.%, to 1594,1-3	3	1	8*	t

Table 1:  Starred entries are Shakespeare rejections; unlike 2H6 and R3, 
Woodstock has far too many to be a Shakespeare could-be. In our regular 
48 tests, Woodstock falls outside Shakespeare's 1590's range 20 times, 
2H6 thrice, Richard III never.  By contrast, none of our 28 other core 
Shakespeare baseline plays have more than two rejections. 13 have one 
rejection; 9, like R3, have none.  Midline speech endings are not part 
of our regular test suite, but Woodstock also has twice as many of them 
as any Shakespeare play written before 1594.  Source: our 2004, pp. 
398-420. Remarks:  * = rejection, sample falls outside Shakespeare 
profile. g = test is sensitive to genre; p = test is sensitive to 
prosody; e = test is sensitive to editing; t = test is sensitive to time 
of composition.  For a description of each test, see our 1996, 196-203.

These calculations, as indicated, presuppose that Mr. Egan's dating of 
Woodstock, and the Riverside Shakespeare's of the other two, are 
correct. They also presuppose that the Louis Ule/Tucker Brooke edition 
of Woodstock we used is roughly consistent in its editing with that of 
the Riverside Shakespeare, which we used for our Shakespeare baseline. 
Mr. Davis correctly observes that some of Shakespeare's stylistic habits 
changed over his writing lifetime.  We believe that eight of our regular 
48 tests are sensitive to time of composition, and that five of these 
are involved in our time-capsule comparison, the ones marked "t" in the 
remarks column.  If, as MacDonald Jackson argues and we agree, the play 
was more likely written in the 1600's, most of these rejections, maybe 
all of them, would disappear.  But the central pillar of Mr. Egan's 
arguments that Woodstock and Richard II were a sequentially written, 
dovetailed pair would disappear with it.  Moreover, the play would still 
have 15 rejections and, hence, still have astronomically low odds of 
Shakespeare authorship.

Four more Woodstock-rejecting tests, marked "e" in the remarks column, 
can be sensitive to the way the text was edited and may not reflect 
actual differences between one author and another.  We have not tried to 
find whether this is actually so of our Tucker-Brooke Woodstock edition 
because it's a lot of work, for which others are better suited, and 
because it would make little difference in the final result.  Even if 
all four editing-sensitive tests were ignored, along with all five 
time-sensitive tests, Woodstock would still have 11 rejections and would 
still test in a different galaxy from any play in our Shakespeare 
baseline.

Stylometry, in essence, is a way of trying to predict what we don't know 
from what we do know. We know from years of testing that Shakespeare's 
known, single-authored plays and poems are remarkably predictable. 
Without exception, they all fall within a narrow range of composite 
probability, and they continue to do so, as we have just shown with R3, 
if you pull one out and test it against all the rest, and with no known 
false positives or false negatives. See our 2004, pp. 357, 365-368.

Where the sample text is short, or possibly co-authored, or written 
before or after Shakespeare's writing lifetime, we cannot speak with 
such assurance, and we have made a special effort in recent years to 
tell our readers not just whether we think a text is or is not 
Shakespeare's but how well we think we know it and how much it differs 
from Shakespeare.  In our SHAKSPER postings over the years we have 
reported many close calls, such as the "Shakespeare sections" of Edward 
III and Sir Thomas More, and some middling-close calls, such as A 
Lover's Complaint. 2H6, though technically just outside our ballpark, is 
a very close call and would fall within the ballpark had we defined it 
one rejection more loosely or had drawn one of our profiles a point 
wider.  Both of these would be perfectly reasonable safety allowances 
for latent variability (our 2004, p. 366), and we suppose that 2H6, 
though not in our gold-standard baseline, could easily be Shakespeare's.

But Woodstock is not a close call.  It is more like The Funeral Elegy 
and the poems of Oxford, Bacon, and Marlowe, where the works in question 
tested in a different galaxy from our Shakespeare baseline.  If 
Woodstock turned up in a time capsule along with R3 and 2H6, it would be 
not be at all hard to tell which of the three "very early" plays could 
be Shakespeare and which could not.  We might not go so far as the path 
breaking Woodstock scholar, A.P. Rossiter, in saying "There is not the 
smallest chance that he [Woodstock's anonymous author] was Shakespeare 
(Rossiter, 1946, p. 73)," but it does seem to us that the odds of common 
authorship with Shakespeare are much lower than the odds of getting hit 
by lightning or of winning the lottery.  The same cannot be said of 
either of our very early Shakespeare plays, the gold-standard Richard 
III, and the open-to-question Henry VI, Part II.  Early though they are, 
they seem to us as clear, easy Shakespeare could-be's by our tests as 
Woodstock, whether early or late, is a clear, easy couldn't-be.  If Mr. 
Egan had accepted our challenge and let the question be put to a vote, I 
would be surprised if SHAKSPER's membership thought otherwise.

References:

Elliott, W. E. Y. and R. Valenza (2004). "Oxford by the Numbers: What 
are the Odds that the Earl of Oxford Could Have Written Shakespeare's 
Poems and Plays?" Tennessee Law Review 72(1): 323-454. 
http://govt.claremontmckenna.edu/welliott/UTConference/Oxford_by_Numbers
.pdf

Elliott, W. E. Y. and R. J. Valenza (1996). "And Then There Were None: 
Winnowing the Shakespeare Claimants." Computers and the Humanities 30: 
191-245.

Jackson, M. P. (2002). "Shakespeare's Richard II and the Anonymous 
Thomas of Woodstock." Medieval and Renaissance Drama in England 14: 17-65.

Rossiter, A. P., Ed. (1946). Woodstock: A Moral History. London, Chatto 
and Windus.

_______________________________________________________________
S H A K S P E R: The Global Shakespeare Discussion List
Hardy M. Cook, editor@shaksper.net
The S H A K S P E R Web Site <http://www.shaksper.net>

DISCLAIMER: Although SHAKSPER is a moderated discussion list, the 
opinions expressed on it are the sole property of the poster, and the 
editor assumes no responsibility for them.


about SHAKSPER | current postings | submitted papers | browse SHAKSPER | search SHAKSPER
  
Copyright © 2004, Hardy M. Cook, design by Eric Luhrs. All rights reserved.