"Shakespeare's Early Career" by Gabriel Egan
Just when and how did Shakespeare start his career as a writer? One kind of evidence is easy to come by, since the appearance of a book in print must follow, not precede, the composition of the words in the book. We may start then with the earliest Shakespeare editions [SLIDE]:
1592 Arden of Faversham (a play)
1593 Venus and Adonis (a poem)
1594 Titus Andronicus (a play)
1594 The Rape of Lucrece (a poem)
1594 The Contention of York and Lancaster / 2 Henry 6 (a play)
1595 Richard Duke of York / 3 Henry 6 (a play)
1596 Edward 3 (a play)
I have omitted from this list the reprints of existing editions: these are just the first editions. I take as accepted the recent arguments for Shakespeare contributing to Arden of Faversham and Edward 3 and that he collaborated with others, quite possibly including Christopher Marlowe, on his first two plays about King Henry 6, and that George Peele wrote the first act of Titus Andronicus (Craig & Kinney 2009; Craig & Burrows 2012). A key fact to observe, then, is that every one of these plays was a collaborative effort by Shakespeare, although he appears to have worked on his own for the poems. On this evidence, early Shakespeare is collaborative Shakespeare.
But of course printed play editions are only one kind of evidence for dating the beginning of Shakespeare's career. We also have allusions to Shakespeare as a writer, starting with Robert Greene's in 1592, and in 1598 [SLIDE] Francis Meres helpfully listed Shakespeare's works as the comedies The Two Gentlemen of Verona, The Comedy of Errors, Love's Labour's Lost, Love's Labour's Won, A Midsummer Night's Dream, The Merchant of Venice and the tragedies Richard 2, Richard 3, Henry 4, King John, Titus Andronicus, and Romeo and Juliet. Of these, The Two Gentlemen of Verona, Love's Labour's Won, The Comedy of Errors, A Midsummer Night's Dream, The Merchant of Venice, and King John had not, by 1598, appeared in print. Leaving aside the possibility of editions now lost--certainly true of the first edition of Love's Labour's Lost (Freeman & Grinke 2002) and perhaps true of Love's Labour's Won--we assume that Meres knew these plays from seeing them performed or by reading them in manuscript. Some were soon printed, but The Two Gentlemen of Verona, The Comedy of Errors and King John remained unprinted until 1623. On the evidence we have seen so far, these plays could have been written any time before Meres mentioned them in 1598.
There is other evidence to use, including the registration of manuscript copies at Stationers' Hall, and of course the wider facts of Shakespeare's biography and his 'lost years' of 1585-1592 have to be take account of when thinking about his first efforts as a writer. [SLIDE] Andrew Gurr thought that Sonnet 145 with its pun of Hathaway and hate-away--"'I hate' from hate away she threw, | And saved my life, saying 'not you'"--was written in the summer of 1582 by the eighteen-year old Shakespeare as he wooed Ann Hathaway (Gurr 1971). The reading is poetically convincing, although the dating of 1582 relies on the assumption that Shakespeare ceased writing bad poetry to Ann once they married.
Bad writing has been crucial in dating The Two Gentlemen of Verona as Shakespeare's very first play for the commercial theatre. The Oxford Complete Works of Shakespeare observes that ". . . its dramatic structure is comparatively unambitious, and while some of its scenes are expertly constructed, those involving more than, at the most, four characters betray an uncertainty of technique suggestive of inexperience (Shakespeare 2005, 1). This is essentially a paraphrase of Stanley Wells's influential article entitled "The Failure of The Two Gentlemen of Verona" (Wells 1963). As early as 1725, Alexander Pope referred to this play as "suppos'd to be one of the first he wrote" (Shakespeare 1725, 155). All eighteenth-century commentators blamed Shakespeare's inexperience for the play's problems, except Thomas Hanmer and John Upton who believed The Two Gentlemen of Verona to be largely or wholly non-Shakespearian (Shakespeare 1744, 143n; Upton 1746, 274). Suspicions about the authorship resurfaced in the late-nineteenth and early twentieth centuries (Fleay 1875; Fleay 1876, 27-30; Fleay 1886, 188-91; Sarrazin 1896, 163; Shakespeare 1906; Shakespeare 1921; Robertson 1923, 1-44; Parks 1937; Tannenbaum 1939; Shakespeare 1969, 22-35), connected to a suspicion that the play is stratified in consisting of two layers of writing--perhaps different hands, perhaps one hand at different times--that can be distinguished by the sensitive ear.
I lack the sensitive ear to perceive these strata, but I know how to run the computational stylistic tests devised by John Burrows and Hugh Craig that are demonstrably capable of distinguishing authors by their choices of words (Burrows 2007; Craig & Kinney 2009; Craig & Whipp 2010). Their test called Zeta compares two electronic texts and finds the words that are most common in one and least common in the other, and then counts how often these words occur in a third text--the sample being tested--so we can say which of the first two texts it is more like. To see how this works, let us take a text that everyone agrees is stratified by co-authorship. The Two Noble Kinsmen has 2918 lines by John Fletcher (call that stratum A) and 2131 lines by Shakespeare (call that stratum B). I divided these strata in 500-line segments, A1-A6 of Fletcher and B1-B4 of Shakespeare. Each experiment requires taking one segment out of each stratum and comparing these two with the strata left behind.
For this first plot [SLIDE], I removed segments A1 and B1 from rest of the A and B segments. Then I generated the lists of words that most distinguish what remains of each stratum, that is the words that distinguish the whole of A2-A6 from the whole of B2-B4. We simply count how many of these words occur in each test segment, A1 and B1. Here the x-axis shows how many of the A-favoured words each segment contains and the y-axis how many of the B-favoured words it contains. Naturally enough, segments A2-A6 (the blue diamonds in the bottom right corner of the plot) score highly for A-favoured words and lowly for B-favoured words and segments B2-B4 (the red squares in the top left corner of the plot) score lowly for A-favoured words and highly for B-favoured words. The separation of these clusters is not of itself evidence that strata A and B are different, since we went looking for precisely the words that would produce this separation, those words that are frequent in one stratum and infrequent in the other. What matters is the scores for the sample segments extracted from these strata, the sample segments A1 and B1. As you can see, A1 (the green triangle) has more of the A-favoured words than B1 (the black cross) has, although it also scores more highly for the B-favoured words, the y-axis We are interested in the separation between the two segments being tested: the greater the gap and the nearer they are to their home strata, the more unalike those two strata are.
This plot [SLIDE] shows the results for A2 and B2, with A2 closer still to the other A segments although B2 again scores rather lowly for B segment words. Likewise for segment A3 and B3 [SLIDE] and the segments A4 and B4 [SLIDE]. The significance of these plots is that the two segments taken out of their strata score differently from one another: they are demonstrably unalike in their word choices. To help visual this significance, we can repeat the experiment for a play, Twelfth Night, in which no-one suspects co-authorship. I divided the play into six 500-line units and arbitrarily called the first, the third and the fifth of these segments the A stratum (A1-A3) and the second, the fourth and the sixth of these segments the B stratum (B1-B3). [SLIDE] Here you can see that the two extracted segments A1 and B1 are hardly separated at all: they have about the same number of words characteristic of stratum A as words characteristic of stratum B. [SLIDE] Likewise with segments A2 and B2 [SLIDE] and even more so with segments A3 and B3 [SLIDE]. In short: separated segments show stratified texts, close segments show unstratified texts.
So, to The Two Gentlemen of Verona. For the second Arden edition of the play, Clifford Leech proposed two strata, both by Shakespeare: the initial writing in 1592, our stratum A, and the later writing in 1593, our stratum B consisting of all of Lance's role (2.3, 2.5, 3.1.260-372 and 4.4.1-59) plus scenes 1.1, 1.2, 1.3, 2.2, 2.6 and 2.7. These divide neatly into segments of 1625 words, A1-A6 and B1-B4. As before, we take out pairs A1 and B1, A2 and B2, and so on, and see how alike they are in the usage of words distinctive of the stratum they were extracted from. [SLIDE] Here are the results for A1 and B1, [SLIDE] A2 and B2, [SLIDE] A3 and B3, and [SLIDE] A4 and B4. There is clear separation of the paired segments except for the pair A3-B3. It is the kind of separation we saw with the co-authored play The Two Noble Kinsmen and quite unlike the pattern of closeness we saw for the sole-authored Twelfth Night. What could be causing it? Leech did not propose co-authorship, merely stratification due to Shakespeare commencing the play in 1592 and coming back to complete it in 1593. One would not expect Shakespeare's writing choices to change so much in less than two years, so there appears to be at least a prima facie case for the possibility that the play is co-authored.
The most recent complete chronology of Shakespeare's works, given in the 1987 Textual Companion to the Oxford Complete Works, has his early career starting with these plays in this order:
The Two Gentlemen of Verona (1590-91)
The Taming of the Shrew (1590-91)
Contention of York and Lancaster / 2 Henry 6 (1591)
Richard Duke of York / 3 Henry 6 (1591)
1 Henry 6 (1592)
Titus Andronicus (1592)
(Wells et al. 1987, 109-15)
With the subsequent addition of the co-authored Edward 3 and Arden of Faversham to this early phase of Shakespeare's career and the realization that all three of the Henry 6 plays and Titus Andronicus were co-authored, it is The Two Gentlemen of Verona and The Taming of the Shrew that start to look anomalous as sole-authored plays. Perhaps they were not sole-authored, and Shakespeare really did get his career started by co-authorship, either in active collaborations with other writers or by taking over existing plays and refurbishing them, as a number of eighteen-century commentators long ago suggested.
Burrows, John. 2007. "All the Way Through: Testing for Authorship in Different Frequency Strata." Literary and Linguistic Computing 22. 27-47.
Craig, Hugh and Arthur F. Kinney. 2009. Shakespeare, Computers, and the Mystery of Authorship. Cambridge. Cambridge University Press.
Craig, Hugh and John Burrows. 2012. "A Collaboration About a Collaboration: The Authorship of King Henry VI, Part Three." Collaborative Research in the Digital Humanities: A Volume in Honour of Harold Short, on the Occasion of His 65th Birthday and His Retirement, September 2010. Edited by Marilyn Deegan and Willard McCarty. Farnham. Ashgate. 27-65.
Craig, Hugh and R. Whipp. 2010. "Old Spellings, New Methods: Automated Procedures for Indeterminate Linguistic Data." Literary and Linguistic Computing 25. 37-52.
Fleay, F. G. 1875. "On Certain Plays of Shakspere of Which Portions Were Written at Different Periods of His Life." Transactions of the New Shakspere Society Number 2. 285-317.
Fleay, F. G. 1876. Shakespeare Manual. London. Macmillan.
Fleay, F. G. 1886. A Chronicle History of the Life and Work of William Shakespeare: Player, Poet, and Playmaker. London. John C. Nimmo.
Freeman, Arthur and Paul Grinke. 2002. "Four New Shakespeare Quartos?: Viscount Conway's Lost English Plays." Times Literary Supplement Number 5166 (5 April). 17-18.
Gurr, Andrew. 1971. "Shakespeare's First Poem: Sonnet 145." Essays in Criticism 21. 221-26.
Parks, George B. 1937. "The Development of The Two Gentlemen of Verona." Huntington Library Bulletin 11. 1-11.
Robertson, J. M. 1923. The Shakespeare Canon. Vol. 2: The Two Gentlemen of Verona; Richard II; The Comedy of Errors; Measure for Measure. 5 vols. London. Routledge.
Sarrazin, Gregor. 1896. "Zur Chronologie Von Shakespeare's Dichtungen." Shakespeare Jahrbuch 32. 149-81.
Shakespeare, William. 1725. The Works. Ed. Alexander Pope. Vol. 1: Preface; The Tempest; A Midsummer Night's Dream; The Two Gentlemen of Verona; The Merry Wives of Windsor; Measure for Measure; The Comedy of Errors; Much Ado About Nothing. 6 vols. London. Jacob Tonson.
Shakespeare, William. 1744. The Works. Ed. Thomas Hanmer. Vol. 1: The Tempest, A Midsummer Night's Dream, The Two Gentlemen of Verona, The Merry Wives of Windsor, Measure for Measure, The Comedy of Errors, Much Ado About Nothing. 6 vols. Oxford. The Theatre.
Shakespeare, William. 1906. The Two Gentlemen of Verona. Ed. R. Warwick Bond. The Arden Shakespeare. London. Methuen.
Shakespeare, William. 1921. The Two Gentlemen of Verona. Ed. Arthur Quiller-Couch and John Dover Wilson. The New Shakespeare. Cambridge. Cambridge University Press.
Shakespeare, William. 1969. The Two Gentlemen of Verona. Ed. Clifford Leech. The Arden Shakespeare. London. Methuen.
Shakespeare, William. 2005. William Shakespeare: The Complete Works. Ed. Stanley Wells, Gary Taylor, John Jowett, and William Montgomery. Second edition. Oxford. Clarendon Press.
Tannenbaum, Samuel A. 1939. The New Cambridge Shakespeare and The Two Gentlemen of Verona. New York. Tenny.
Upton, John. 1746. Critical Observations on Shakespeare. London. G. Hawkins.
Wells, Stanley, Gary Taylor, John Jowett and William Montgomery. 1987. William Shakespeare: A Textual Companion. Oxford. Oxford University Press.
Wells, Stanley. 1963. "The Failure of The Two Gentlemen of Verona." Shakespeare Jahrbuch 99. 161-73.