Oxford University Press's
Academic Insights for the Thinking World

What did Shakespeare write?

We have always known already that Shakespeare was a collaborator; he was a man of the theatre which is inherently a very collaborative, social art. The news is that he also collaborated as a writer much more than we used to think he did. We can now say with a high degree of certainty that upward of third of his plays were co-written in some sense or other.

In most film portrayals, Shakespeare seems to produce his plays in isolation, with the works torn from his soul and arising from his own personal experiences alone. When we realise he worked collaboratively, we see him quite differently –  in a community that works together not only in putting the plays on, but also in writing them.

Evidence comes from a number of sources. Firstly, there is published evidence: the 1634 edition of The Two Noble Kinsmen states on its title page that the play was written ‘by William Shakespeare and John Fletcher.’

Secondly, we know the method was common practice at the time: we have very good records for the rival company, The Admiral’s Men, who used freelance writers working in teams to get a play done quite quickly.

Thirdly, we can tell from the text itself. In the mid-nineteenth century, scholarship examining the minutiae of the writing began to throw up the suspicion that some of the plays were in fact co-written. The main method used has been comparing the fine detail of different writers’ styles, by measuring the frequency of particular features, be it the frequency of certain words, or poetic traits such as feminine endings to verse lines, or certain kinds of prose style.

The things that were counted were things that was easy for a human to count. But that could only take the research so far. In the last 20 or 30 years this approach has been greatly enhanced by the availability of machines to do the counting. Whilst previous generations of researchers were assiduous in the work they did, they couldn’t aim at a comprehensive approach – no-one has the time to study all of the verse features or all of the rhyme schemes found in all of the writing of Shakespeare’s time.

Untitled by Eflon. CC-BY-2.0 via Flickr.
Untitled by Eflon. CC-BY-2.0 via Flickr.

And it’s that kind of comprehensive approach that really enables one to examine Shakespeare in the context of a group of his contemporaries. For this type of analysis, we don’t want to treat him as ‘special’; we want to see if his style emerges from the other writers of the period as something distinctive that we can measure. To do that, we need to have a substantial body of the writing of the period available in digital form, and that has become true in the last 10 or 20 years.

The new machine-based approach – Computational Stylistics – has started to reveal some very startling facts. For example, it is now clear that Shakespeare’s vocabulary – the total body of all the different words he knew – was not exceptionally large (as has long been assumed) but rather was just typical for the period. We now know that a lot of words and phrases that we used to think were coined by Shakespeare were already in use by other writers before him. Wherever his genius lay, it was not in his vocabulary, but in his ways of combining existing words and phrases.

The computational approaches must be used with care, of course. The history of authorship attribution is full of cases of self-delusion in which an investigator discovered a new method, applied it, found that it seemed to confirm some existing beliefs about the text, and then the applied it elsewhere to produce a new set of ‘startling discoveries’ which later turned out not to be true. The worst case of this is Donald Foster’s SHAXICON database that was supposed to correlate Shakespeare’s roles as a performer with his choice of words for his next play after acting each role. It’s an interesting hypothesis – that learning a particular role for performance put certain words to the front of Shakespeare’s mind – but Foster was unable to prove it, and no-one else was able to replicate his claims, and the one attribution that Foster made by his methods, that A Funeral Elegy to William Peter is by Shakespeare, was soon disproved.

When addressing a particular question, the truly scientific approach is to have several independent studies, using different methods, working on the same texts. If they all come to the same conclusion by different methods and we cannot figure out how that might happen without the conclusion being true, then we can start to treat the conclusion as highly likely to be true. But if the studies do not agree, the question will become ‘why not?’ Has the discrepancy revealed unacknowledged investigator bias or something faulty with the methods? This is how science works: we have hypotheses, we devise experiments to test them, and we base our beliefs on the most plausible explanations that account for the results that we find. We should apply this kind of scientific rigour as much to humanistic study as anything else, since no matter what their fields everyone who undertakes research for a living is ultimately in pursuit of the truth, and these are the best ways we have for finding it.

Featured image credit: Front cover of William Shakespeare’s First Folio by Ben Sutherland. CC-BY-2.0 via Flickr.

Recent Comments

  1. Richard M. Waugaman, M.D.

    There is another perspective on the co-authorship of Shakespeare’s works, not mentioned by Gabriel Egan. It flows from a different authorship paradigm—one accepted by those of us whom Gary Taylor has unapologetically compared with Holocaust deniers. We endorse the 1920 theory that Edward de Vere, Earl of Oxford, used the pen name “Shake-speare.” De Vere’s 1570 Geneva Bible is at the Folger Shakespeare Library. Its handwritten annotations show a close correlation with Biblical echoes in Shakespeare. Its musical translation of the Psalms, bound at the back, have revealed the largest literary sources from the Psalms for the works of Shakespeare, discovered because of the 14 psalms marked with large, hand-drawn “manicules,” or pointing hands.

    De Vere had a close association with several possible literary collaborators. For example, his uncle Arthur Golding is said to have translated Ovid’s Metamorphoses when he and the adolescent de Vere were living in the same household. That translation was called “the most beautiful book in the [English] language” by Ezra Pound. It is known to be one of the top literary sources for Shakespeare (along with the Bible and Plutarch). It shows a rich pattern of hendiadys—the verbal doubling so beloved by Shakespeare (“slings and arrows,” “sound and fury,” etc.). This rhetorical form was heavily used by Virgil, but no so much by de Vere’s contemporaries. Spelling quirks also suggest that de Vere, not his uncle, may have been the real translator.

    The Elizabethan playwright who was most successful in having his plays staged at court during the decade after the Queen established censorship of court entertainments by the Master of Revels was John Lyly. He set his plays in ancient times to evade concerns about commentary about current events. It is not well known that he was employed as de Vere’s literary secretary at the time, again suggesting possible collaboration.

    Anthony Munday was a prolific Elizabethan author, who wrote the play Sir Thomas More that is said to show the handwriting of Shakespeare (“Hand D”) in one scene. Again, Munday worked for de Vere, and the spelling quirks in Hand D match those of de Vere’s extant letters.

    People often ask why de Vere would have concealed his authorship. That was the norm for earls in his day (with infrequent exceptions). We have a letter from the Earl of Essex’s secretary that illustrates how an earl tried to conceal his own writings. Essex wrote a self-serving account of his role in the 1596 battle at the Spanish port of Cadiz. He asked that his account be signed “R.B.” His secretary wrote, “which some noe doubt will interpret to be Mr. Beale.” Similarly, de Vere may have hidden his own literary works behind the “allonyms” of William Shakspere of Stratford, as well as behind Golding, Lyly, and Munday.

  2. A. Crampin

    It is crucial in this work that everybody makes crystal clear exactly what evidence and methodology they are relying on, and what assumptions have been made before they draw any inferences. This must happen every step of the way.

    Dr Waugaman was careful about the Thomas More manuscript. Others are much less so.

    Taking for granted is the curse of Shakespeare scholarship.

  3. Gary Goldstein

    I’d like to make two points. John Drakakis of the University of Stirling told Constance Grady of Vox.com that, “Dramatists could imitate each other, so what looks like a Marlovian style might have been Shakespeare imitating someone else.”

    Grady added that it’s also not clear that OUP can definitively say that Marlowe used the phrase “glory droopeth” more than any of his contemporaries. Of all the printed matter produced in the Elizabeth era, only about 15 to 20 percent survives today, Drakakis told her. That means that it’s much more difficult to make a statistically significant claim about the relative frequency of Elizabethan word usage than Taylor and his colleagues are suggesting.

  4. Mike Leadbetter

    If you are using 20% of something as large as the early modern professional theatre, you have more than enough data to make statistical claims, especially when comparing front rank playwrights since the better the author, the more work has survived.

    Modern stylometry has developed well enough to both distinguish the collaboration of a number of authors in Shakespeare’s work and to begin to detect Shakespeare’s work in plays not attributed to him. It has long since been sufficiently sophisticated to eliminate that claims of impostors, such as Edward de Vere, which are, in any case, entirely unsupported by evidence.

    It s true that clarity is required concerning the methodology of stylometric claims. It is not true that Dr Waugaman was careful in his reference to British Library-MS. Harley 7368, the incomplete manuscript of Sir Thomas More which has additions by Shakespeare which are now accepted as canonical by everyone except a few contrarians. Palaeographers are satisfied the hand is Shakespeare’s and the work is certainly good enough. It’s existence demolishes theories which advance a hidden, allonymous author.

    The ‘different authorship paradigm’ is the Looking Glass World of Oxfordianism—a pseudo-academic game show where facts are ignored, attribution is a guesswork, words mean exactly what you want them to mean, and it’s always time for tea.

  5. Tim Backer

    Shake- speare might have been his formal role, and It’s doubtful that many scholars would look into a possible military or intelligence role. We know that the plays reveal Court political matters otherwise unknown. For all we know, his first best woman was

  6. Richard M. Waugaman, M.D.

    For more on the ways that the “Oxfordian” authorship theory is entering the mainstream, please see

Comments are closed.