X hits on this document

26 views

0 shares

0 downloads

0 comments

3 / 13

the only ’suspect’ we are investigating to have lived throughout the entire lifetime of William Shakespeare [7]. Works by Francis Bacon we use for comparison to Shakespeare’s plays include, The Great Instaura- tion, Preparative toward a Natural and Experimental History, and New Atlantis [8].

4.1

Character

Distribution

We used the same character distribution test as with Marlowe. The results are given in Figure[3] with the

2 test displayed in Table[4]. A quick look confirms gives us the same conclusion as with Marlowe. Bacon seems to use longer words (ie fewer spaces) than Shakespeare as well as significantly more useages of the letters ”t”, ”i”, and ”e”.

It’s not surprising that the P-value = 0 for the test given the di erences illustrated by Figure[3]. That means based on this one test, it’s unlikely the works attributed to Bacon and Shakespeare came from the same source. However this result also su ers from the same flaw as in the previous section. Marlow did not have many published works so our character probabities are only based on the three we could find. 2

4.2

Word Length Analysis

The comparison of word length distribution between Shakespeare and Bacon produced significantly dif- ferent results compared to the Marlowe comparison. The graph of the word length distributions can be found in Figure[4].

The shown

P-value from our

2

in

Table[5]

is

2.25E-7.

test for

This

is

independence an extremely

low P-value and leads us to conclude length distribution is significantly di

that the word erent between

Shakespeare and Bacon. From this result, and ers in our project, we decided that it is unlikely

oth- that

Francis

Bacon

ever

wrote

under

Shakespeare’s

name.

4.3

Proportion of Unique Words

Analyzing Bacon’s unique word usage compared to- tal word usage yielded very similar results to that of the Marlowe analysis as shown in Table[3]. Bacon’s average ratio was 0.2044 with a slightly greater value of 0.0012. However, these values are still significantly far away from Shakespeare’s 0.16 average ratio with a variance of 0.0002.

5

Edward de Vere 17th Earl of Oxford

The final Shakespeare candidate we investigate is Ed- ward de Vere. Edward de Vere was born in 1550 to the Earl of Oxford, John de Vere. Edward was edu- cated at Queen’s College Cambridge and also studied law at Gray’s Inn. At a young age, de Vere trav- eled around Europe, visiting France, Germany, and Italy [4]. Oxfordians, those who believe Edward de Vere wrote Shakespeare’s plays, point to these trav- els as evidence of de Vere’s potential for writing plays which show worldly education, as well as an accute interest in things Italian. Other evidence often at- tested to is the amount of literary praise given to de Vere, despite only a few number of his poems surviv- ing [5]. It is these poems that we compare to William Shakespeare’s poems.

5.1

Character

Distribution

Our test for character distribution is identical to that used in the previous sections. There is a di erence though in our corpus of data: Edward de Vere was a noble man and thus it is conjectured that he did not publish plays (such an act was not considered sightly for a man of his status). Thus the only readly available works we could find were 25 attributed short poems for a total of around 600 lines. Because po- etry tends to have a di erent structure and style then prose (ie plays), we did not deem it fair to compare de Vere’s poems with Shakespeares plays. The in- put to our character distribution test is then a col-

lection of Shakespeare’s poems and around 1100 lines. The results are

with Figure[5 plotting the relative

sonnets totaling presented below probabilites and

Table[6]

showing

the

2

test.

There are significant di erence in the character us- age between the two poets. Shakespeare tends to uses fewer spaces (and thus longer words) than does de Vere, and seems to use a lot more punctuation.

The

2

test

shows

fairly

conslusively

that

the

po-

ems attributed to de Vere and Shakespeare came from

di

erent

sources

based

on

the

character

distribution

test. But this test su ers the most in size. We only had some 600 lines of Vere; compare that to the 9000 lines

terms of corpus poetry from de of prose (which

tends to and you

be longer) which we see the heart of the

analyzed

for Marlowe

problem.

Until more

surviving works from de Vere surface, there’s nothing we can do to improve accuracy.

really

3

Document info
Document views26
Page views26
Page last viewedTue Dec 06 20:19:56 UTC 2016
Pages13
Paragraphs972
Words2951

Comments