 | |  |
Notes
Thomas Ravenscroft, Titus Andronicus, or the Rape, of Lavinia . . . A Tragedy Alter'd from Mr Shakespears Works (1687), address to the reader; reprinted in Shakespeare: The Critical Heritage: Volume 1 1623 — 1692, ed. Brian Vickers (1974), p. 239.
The Plays of William Shakespeare, ed. Samuel Johnson (1765); reprinted in Vickers, Critical Heritage: Volume 5 1765 — 1774, p. 142.
Roland Barthes, "The Death of the Author," Image-Music-Text, trans. Stephen Heath (1977), pp. 145 — 147.
Francis Meres, Palladis Tamia: Wit's Treasury (1598); reprinted in E. K. Chambers, William Shakespeare: A Study of Facts and Problems (1930), 2, 193 — 195.
There is a very full discussion of Q Titus Andronicus as a "foul papers" text by Stanley Wells in his Re-Editing Shakespeare for the Modern Reader (1984), pp. 79 — 125.
The specified period is from the opening of the first London public theatre (The Theatre) to the closing of all theatres by the Puritans. My confidence about the rarity of the "others as many as can/may be" formula derives from the presence at Tokyo Gakugei University of an electronic data-base into which stage directions from almost all available plays listed in the Schoenbaum-Harbage Annals have been entered. I am extremely grateful to Professor Yasumada Okamoto and to Mr Yukio Kato for supplying me with relevant information. At the time when they ran computer checks for me, they had entered 508 of the 583 plays of 1576 — 1642 that are listed in the chronological table in The Cambridge Companion to English Renaissance Drama, ed. A. R. Braunmuller and Michael Hattaway (1990), pp. 419 — 446, which omits entertainments, pageants, masques, school plays, and Latin academic plays. I have made some attempt to check plays not yet in the data-base. A further 133 plays outside my chronological limits had been entered, and none exhibited the Peelean formula.
I have worked with The Norton Facsimile: The First Folio of Shakespeare, ed. Charlton Hinman (1968); Shakespeare's Plays in Quarto, ed. Michael J. B. Allen and Kenneth Muir (1981); and Marvin Spevack, A Complete and Systematic Concordance to the Works of Shakespeare: Volume 7: Stage Directions and Speech-Prefixes (1975). References to the Folio (F) and the quartos (Q) or octavo (O) are by signature. Act and scene numbers are those of The Riverside Shakespeare, ed. G. blakemore Evans (1974).
The verb "say" and its inflexions can be checked in Spevack's Concordance. Seeming instances turn out to be references to Lord Say.
For Peele's plays I have used copies of the original quartos in the Bodleian Library, Oxford, but in interpreting stage directions and speech prefixes and determining where prefixes have not been used I have been guided by the four-volume Yale edition of The Life and Works of George Peele, general ed. Charles Tyler Prouty (1952 — 1970).
There are also two or three directions ending in ". . . speaketh" but followed by a speech heading.
Stanley Wells and Gary Taylor with John Jowett and William Montgomery, William Shakespeare: A Textual Companion (1987), p. 210 (note on 1.1.17.1).
J. C. Maxwell, "Peele and Shakespeare: A Stylometric Test," Journal of English and Germanic Philology, 49 (1950), 557 — 561; summarized in Maxwell's Arden edition, pp. xxx — xxxi.
R. F. Hill, "The Composition of Titus Andronicus," Shakespeare Survey, 10 (1957), 60 — 70. Hill draws on the much fuller treatment in his Oxford B.Litt. thesis (1954), "Shakespeare's Use of Formal Rhetoric in his Early Plays up to 1596."
Baron Brainerd, "The Chronology of Shakespeare's Plays: A Statistical Study," Computers and the Humanities, 14 (1980), 221 — 230.
The other "deviant" plays were the three Parts of Henry VI, Pericles, and Love's Labour's Lost. Questions of revision or plural authorship in these plays have not, of course, been settled; John Kerrigan, "Love's Labour's Lost and Shakespearean Revision," Shakespeare Quarterly, 33 (1982), 337 — 339, argues against revision in Love's Labour's Lost.
G. Sarrazin, "Wortechos bei Shakespeare," Shakespeare Jahrbuch, 33 (1897), 121 — 165, and 34 (1898), 119 — 169.
When a chi-square test is used to compare the observed figures 33:25 and 37:68 with those expected on a purely proportional basis, the result is chi-square = 7.04, 1 degree of freedom (or 6.31, 1 d.f., with Yates's correction). There is a description of the chi-square test in Anthony Kenny, The Computation of Style: An Introduction to Statistics for Students of Literature and the Humanities (1982), pp. 110 — 119. Of course, strictly speaking, the test is valid only if the prediction has been made in advance that Part A will differ from Part B in respect of the proportion of links with the first group of plays. A chi-square test that simply compares the two sets of four figures (37:26:19:23 and 33:9:8:8) yields a result that falls short of statistical significance: chi-square = 7.3, 3 d.f.
M. W. A. Smith and Hugh Calvert, "Word-Links as a General Indicator of Chronology of Composition," Notes and Queries, 234 (1989), 338 — 341. Smith and Calvert worked with words found up to ten times in Shakespeare's dramatic canon, as did Eliot Slater in his research into vocabulary and chronology.
Marina Tarlinskaja, Shakespeare's Verse: Iambic Pentameter and the Poet's Idiosyncrasies (1987), p. 124. Tarlinskaja divides Titus Andronicus into "Group 1" scenes and "Group 2" scenes. Her Group 1 is identical with my Part A, except that I omitted the very short 2.2. I explore the implications of her findings in "Another Metrical Index for Shakespeare's Plays: Evidence for Chronology and Authorship," Neuphilologische Mitteilungen, 95 (1994), 453 — 458.
See Textual Companion, pp. 80 — 89. The most rigorous research on high-frequency words in English Renaissance drama has been reported by M. W. A. Smith in a series of articles in Computers and the Humanities, The Journal of Literary and Linguistic Computing, Notes and Queries, and elsewhere, beginning in the early 1980s. Smith's "Forensic Stylometry: A Theoretical Basis for Further Developments of Practical Methods," Journal of the Forensic Science Society, 29 (1989), 15 — 33, provides a mathematical validation of his procedures.
The odds are reckoned as follows: We can form a 2 x 2 contingency table of (a) occurrences of and and (b) occurrences of other words in (c) Act 1 and (d) Acts 2 — 5, where "a" and "b" are the rows and "c" and "d" the columns, and apply a chi-square test, using Yates's correction. Chi-square = 11.273 for 1 degree of freedom, p < 0.001. For 2.1 the rate of and is also very high, 4.512; for 4.1 it is normal, but there is certainly a highly significant disparity between Parts A and B in toto, as originally defined.
For 2.1 the rate is 1.105 and for 4.1 it is 0.972, so that a notable disparity would remain if we were to revert to our original lines of demarcation between Parts A and B.
Working from the modern-spelling edition, The Dramatic and Poetical Works of Robert Greene and George Peele, ed. Alexander Dyce (1874), I keyed text into the computer until approximately two thousand words had been accumulated and a scene end had been reached. The analysis was carried out by means of the Micro-OCP concordance programme.
Figures are based on Maxwell's Arden edition for Titus Andronicus, on the Riverside Shakespeare for other Shakespeare plays, and on Dyce's Works for Peele's plays.
The calculation is based on chi-square testing of a 2 x 2 contingency table of (a) iambic pentameter lines ending with the "of my right" formula and (b) iambic pentameter lines not ending with the formula in (c) Act 1 and (d) the rest of the play.
Since this article was completed confirmation of its findings has come from two sources. "The Shakespeare Clinic" run by Ward Elliott and Robert Valenza of Claremont McKenna College has released a report entitled "Matching Shakespeare, 1994: Computer Testing of Elizabethan Texts for Common Authorship with Shakespeare" (June 21, 1994; and Supplement, July 15, 1994). Elliott and his associates have applied a battery of tests to texts by Shakespeare and by a fair range of his contemporaries. They show that all the plays traditionally accepted into the Shakespeare canon pass as Shakespearean on all or all-but-one of nineteen computerized tests — that is, the plays are consistent with one another, within statistically acceptable limits — except for those considered collaborations by the Oxford Shakespeare (I Henry VI, Timon of Athens, Pericles, Henry VIII, and The Two Noble Kinsmen) and Titus Andronicus and 3 Henry VI, each of which is recognized by the Oxford Textual Companion as possibly collaborative. The failure of Titus to meet the Shakespearean norms is particularly marked. The play as a whole is rejected by five tests, and a stratum that is labelled (probably misleadingly) "early," that includes Act 1, and that roughly corresponds to my "Part A" is largely responsible for the anomalous results: it is rejected by no fewer than seven tests, while the other stratum is rejected by only two. Further, in "Common Words in Titus Andronicus: The Presence of Peele," forthcoming in Notes and Queries, Brian Boyd re-examines and quantifies the evidence of formulaic repetition of common words in the suspect parts of Titus and in Peele, and assigns to Peele the bulk of 1.1 — 2.2 and 4.1.
| | |