Frequency Analysis
New ways to deconstruct De Vere digitally deprive Doubters of dispositive data
In the same way that they welcomed digital stylometry, Doubters were keenly interested in the arrival of large, organised corpora of digital texts. They presented new research tools and new ways to count things and calculate probabilities. Counting function words — the words writers use unconsciously — could be used as a recipe for analysing style, but once again large data turned out not to be helpful in promoting one writer’s claim to the corpus of another.
Counting the usage of individual words and searching for tell-tale expressions, however, became much easier once a single search could extract results across an entire corpus. This also went wrong when Roger Stritmatter wrote a whole book attempting to connect De Vere and Shakespeare by counting unusual or rare expressions — miscounting, in fact.
EEBO V3 is the latest collection of Early English Books Online, though the long awaited EEBO has been only “a few weeks away” for almost three years now. V3 differs from previous versions in that you can count parts of speech as well as identify function words. You can judge the usefulness of this for yourself on the site maintained by Lancaster University and see just how quickly you can separate one poet from another.
If you want to work with Oxford, a resources section contains a De Vere corpus we created which can be uploaded for comparison with their Bankside corpora, all of which have been hand-keyed and inspected by the team at Lancaster University. Our De Vere corpus, while not exactly hand-keyed, contains the 20 poems claimed as canonical by the De Vere Society and his 45,000-word letter repository which you can browse on this site. It too has every word tagged with metadata.
The nature of stylistic enquiry has completely changed. We used EEBO V3 to extract haggard hawks but the new version lets you search for haggard as both noun and adjective and pair it with other birds or other nouns, categorising your search — in the case of Shakespeare — to plays, genres, even individual characters or characters grouped by their age or social status. You can now analyse the difference between aristocrats, the middle class and the groundlings to see the variations Shakespeare used in vocabularies and different parts of speech.
This is a new field of study and so unlikely to assist Doubters with anything other than final enlightenment.
Alliteration
If you’ve read Oxford’s poetry you will have been struck by his use of clangorous alliterative lists of nouns and adjectives.
My life, through ling’ring long, is lodg’d in lair of loathsome ways;
Now this really doesn’t sound like Shakespeare at all, except when he is mocking over-reaching artistic pretensions. There are more instances, however, of Shakespeare alliterating multiple words than you might at first think. The three tables below cover his long poems and sonnets, the First Folio plays, and a selection of contemporary dramatists for comparison. Oxford comes last — his entire surviving poetic output, machine-tagged from a single corpus. Spend two minutes with the Shakespeare tables first.
DIY EEBO research
Rather than write another lecture on stylometry, check out EEBO for yourslef. These tables extract an alliterative pair, an afjective and a noun, and pull out any other alliterative content nearby. There are four, three available as standard at The University of Lancaster’s implementation of CQP, a very usable large language search tool. The fourth is an Oxford corpus built by ourselves which can be uploaded and used in a Lancaster CQP account and can be downloaded from our Resource page.
The tables can be sorted, filtered an searched. If you search the First Folio table for “Midsummer” you can compare Bottom’s alliterative fourteeners to Oxford’s. If you compare the alliteration in Macbeth to that in Much Ado about nothing you will see how far Shakespeare adapts different writing styles and vocabulary to different themes and circumstances. Nobody seriously belives Benedict is going to kill Claudio but the language itself tells us what fate lies in store for Macbeth before he gets home in Act I.
By sorting alphabetically you can see, the same alliterative pairing used differently Venus and Adonis and Lucrece, or see how Marlowe and Jonson handle the same device.
There are almost
Shakespeare’s poems and sonnets
Shakespeare’s plays (First Folio)
Contemporary dramatists
Edward de Vere, 17th Earl of Oxford — Juvenilia and letters
The table below covers de Vere’s entire surviving poetic output — the 20 poems accepted as canonical by the De Vere Society — together with his letter corpus. The poems were written largely in his teens and twenties; the letters span his adult life. Unlike the Shakespeare tables above, which are drawn from hand-keyed, editorially verified corpora, the Oxford corpus was processed by treetagging software, so the data is less refined. The difference in volume is not an artefact of the method. Neither is the difference in quality.
79 instances in Shakespeare’s poems and sonnets | 901 in the First Folio | 1037 across 21 contemporary dramatists | 34 in the complete De Vere corpus (18 from poems, 16 from letters)