Notes and Queries, March 1997 v44 n1 p75(7)
Phrase lengths in 'Henry VIII' Shakespeare and Fletcher.
Abstract: Scholars have attributed the play Henry VIII to English dramatists William Shakespeare and John Fletcher using metrical information, verbal similarities and idiosyncracies of phrasing to back up their theory. An approach using phrase lengths to measure scenes supposedly written by Shakespeare and those said to be written by Fletcher seems to support a theoretical division of the play between the two dramatists.
Full Text: COPYRIGHT 1997 Oxford University Press
Responding to a hint from Tennyson that some of the verse in Henry VIII sounded more like John Fletcher's than William Shakespeare's, James Spedding in 1850 produced metrical data in support of a division of the play between the two dramatists. Another Victorian scholar, Samuel Hickson, immediately announced that he had independently established similar allocations on the evidence of verbal parallels and quirks of phrasing. The theory that Shakespeare had written I.i, I.ii, II.iii, II.iv, III.ii.1-203, and V.i, and Fletcher the rest of the play, including the epilogue and prologue, eventually assumed the status of orthodoxy. However, some twentieth-century commentators, including the major critic G. Wilson Knight, continued to ascribe Henry VIII, first published in the Shakespeare First Folio of 1623, to Shakespeare alone, and the view that the play was wholly, or almost wholly, his was promoted in R. A. Foakes's influential new Arden edition (London, 1957).(1)
Then on the basis of 'linguistic evidence' - the pattern of use of such alternatives as ye or you, 'em or them, has or hath, i'th' or in the, and the like - Cyrus Hoy concluded that Fletcher's share of the play was much smaller than Spedding and Hickson had supposed, his role in II.i, II.ii, III.ii.204-458, IV.i, and IV.ii having been limited to the revision of a few sentences and the interpolation of a handful of lines.(2) The New Cambridge Shakespeare editor John Margeson (Cambridge, 1990) thinks that 'one must take Hoy's contribution to the debate as the most carefully worked out evidence to date as to the actual fact of Fletcher's participation and also to its extent' (13).
Gary Taylor provides a brief summary of the current state of the controversy.(3) As he points out, the possibility that there was a third hand in the play complicates assessment. Spedding thought that Beaumont may have been involved with Fletcher in Act IV. Robert Boyle and H. Dugdale Sykes assigned Henry VIII to Fletcher and Massinger.(4) Recent investigators may have been too ready to reduce the plausible authorial candidates to Shakespeare and Fletcher, assuming that lack of clear signs of one within a particular section of the text implies the presence of the other. And yet the consistency with which one objective test after another has confirmed a stylistic split between the 'Shakespeare' and 'Fletcher' shares of the Spedding-Hickson analysis suggests that the traditional allocations are essentially correct.(5) Most tests firmly associate those scenes that Hoy considered basically Shakespearian but slightly revised by Fletcher (II.i, II.ii, III.ii.204-458, IV.i, IV.ii) with the scenes that he regarded as purely Fletcher's (I.iii, I.iv, III.i, V.ii, V.iii, V.iv).
Two careful, mathematically sophisticated recent investigations reach different conclusions, Thomas B. Horton's discriminant analysis of function word rates suggesting that, although Henry VIII must be a collaboration, several scenes should be reassigned from Fletcher to Shakespeare, and Jonathan Hope's 'socio-linguistic' data concerning use of auxiliary 'do' and relative markers strongly supporting the simple binary division between the two dramatists that Spedding and Hickson originally proposed.(6)
There is room, then, for yet another approach to the problem. This note reports on a count of phrase lengths, which seem more suitable data for analysis of authorship in Renaissance plays than the sentence lengths counted by students of authorship in other fields.(7) The textual basis for my calculations is Foakes's Arden edition. Since Foakes argued against collaboration, he is unlikely to have prepared his text in ways that would create purely editorial distinctions between supposedly Shakespearian and supposedly Fletcherian scenes. In particular, he presumably adopted consistent principles of punctuation. This is a matter of some consequence because, for the purposes of this study, length of phrase is defined as the number of words intervening between any of the following marks of punctuation: full stop, question mark, exclamation mark, colon, semi-colon, parenthesis (that is, opening or closing round bracket), dash, and comma. By 'word' is meant a graphic unit preceded by a space. Only dialogue to be spoken on stage has been counted, and the epilogue, prologue, song at the beginning of Act III, and the play's few short patches of prose have been ignored. Data for individual scenes are recorded, in compressed form, in an appendix, while Table 1 highlights the significant results - for the 'Shakespearian', 'Fletcherian', and 'Doubtful' material as a whole.
A chi-square test comparing the overall distribution of phrase lengths for the 'Shakespearian' scenes with that for the 'Fletcherian' scenes reveals a difference that is highly significant, statistically speaking: chi-square = 39.5, 12 degrees of freedom, p [less than] 0.001. Although we cannot assume that literary features will be randomly distributed even throughout a work of single authorship, we seem to have confirmation that those within the portions assigned by Spedding and Hickson to Shakespeare, on the one hand, and to Fletcher, on the other, constitute two distinct populations. When the 'Fletcherian' material is split into that which Hoy considers wholly Fletcher's and that which he considers basically Shakespeare's but with some minor revision by Fletcher, chi-square testing finds them indistinguishable: chi-square = 12.1, 12 d.f., p [greater than] 0.1. In fact random sampling would produce a difference of at least this order almost as often as not. Both subsections of the Spedding-Hickson 'Fletcherian' matter are, in contrast, statistically different from the uncontested 'Shakespearian' matter: comparison of Shakespeare with 'wholly Fletcher' yields chi-square = 39.6, 12 d.f., p [less than] 0.001, and of Shakespeare with 'partially Fletcher' chi-square = 22.5, 12 d.f., p [less than] 0.05. Although the disputed sub-section is the less sharply differentiated from the undoubtedly Shakespearian material, it is nevertheless statistically distinct from it, while closely matching the sub-section ascribed by Hoy to Fletcher alone.
To check the extent of random fluctuation within 'same-author' material, we can divide the 'Shakespearian' scenes into two blocks and the 'Fletcherian' scenes into three, simply according to their numerical order in the play, and test each block against all the others. Roughly equal blocks composed of scenic units may be created as follows: Shakespeare A: I.i, I.ii, II.iii; Shakespeare B: II.iv, III.iia, V.i; Fletcher A: I.iii,I.iv, II.i, II.ii; Fletcher B: III.i, III.iib, IV.i; Fletcher C: IV.ii, V.ii, V.iii, V.iv. The results of the proposed chi-square testing of the distribution of different phrase lengths within pairs of blocks are given in Table 2.
None of the four 'same author' comparisons reveals a statistically significant difference, and nor do the summed results. In contrast, four of the six 'different author' comparisons reveal statistically significant differences, and the summed results have an infinitesimal probability of having been produced by chance. It appears, therefore, that there is acceptable consistency in phrase lengths within the Shakespeare and Fletcher sections of Henry VIII, as these have traditionally been defined, and a definite distinction between those two authorial shares.
Table 1: Phrase Lengths in Henry VIII
(a) (b) (c) (d) (e) (f) (g)
1 124 61 93 154 7.8 6.6
2 154 109 127 236 9.7 10.1
3 196 147 171 318 12.3 13.6
4 208 172 208 380 13.1 16.7
5 157 129 153 282 9.9 12.1
6 157 119 133 252 9.9 10.8
7 126 66 89 155 7.9 6.6
8 114 80 77 157 7.2 6.7
9 116 56 70 126 7.3 5.4
10 74 29 58 87 4.7 3.7
11 43 22 33 55 2.7 2.4
12 26 25 27 52 1.6 2.2
13+ 96 34 49 83 6.0 3.6
Total 1591 1049 1288 2337 100.1 100.5
Column (a) phrase length in words; (b) figures for 'Shakespearian'
scenes (I.i, I.ii, II.iii, II.iv, III.iia, V.i); (c) figures for
scenes attributed by Cyrus Hoy to Fletcher alone (I.iii, I.iv,
III.i, V.ii, V.iii, V.iv); (d) figures for scenes partly Fletcher's,
according to Hoy (Il.i, II.ii III.iib, IV.i IV.ii); (e) figures for
'Fletcherian' scenes according to the traditional Spedding-Hickson
division: i.e. totals of previous two columns; (f) column b, for
'Shakespearian' scenes, as percentages of 'Shakespearian' total; (g)
column e, for 'Fletcherian' scenes according to traditional
division, as percentages of 'Fletcherian' total. Percentages are to
the nearest single decimal place, so fail to add up to 100.
[TABULAR DATA FOR TABLE 2 OMITTED]
One further set of figures makes this very clear. Inspection of the percentages in Table 1 shows that 'Fletcher' uses a higher proportion of phrases of two, three, four, five, six, and twelve words than does 'Shakespeare', who uses the higher proportion of phrases of other lengths. One-word phrases are created mainly by exclamations and vocatives, which neither dramatist could avoid. Otherwise the results are in accord with Fletcher's tendency to build up a passage of verse through a succession of simple grammatical units - by the accretion of 'short appositional phrases by which the thought expands itself in little rushes and eddies'.(8) Shakespeare is more apt than Fletcher to use longer, more complex, phrasing, as sentences straddle line endings in energetic imitation of the tortuous movements of the mind. The overall percentage of phrases of two, three, four, five, six, and twelve words is 65.0 for 'Fletcher', 56.4 for 'Shakespeare'. Table 3 ranks individual scenes according to the percentage of their phrases that are of these lengths. All eleven of the 'Fletcherian' scenes are ranked higher than any of the five 'Shakespearian' scenes. Of course, the percentage for the lowest 'Fletcherian' scene is the merest fraction above that for the highest 'Shakespearian' scene, but the separation is nevertheless perfect.
The lowest ranked 'Fletcherian' scene, II.i, is among those that Hoy reassigns in large part to Shakespeare, but its metrical characteristics and its 'regulation rate' for auxiliary do put it well into Shakespeare's range and outside Fletcher's.(9) The different tests that have been applied to Henry VIII tend to place different scenes within or near the borderland where the two dramatists' practices overlap: no particular scene or scenes consistently, or even repeatedly, yield results that are ambiguous or that contradict Spedding's ascriptions. This would seem to indicate that there really are two authorial styles present in the play and that Spedding's allocations are fairly accurate. Hoy's challenge to the traditional division was based on the fact that in the scenes that he wishes to reassign Fletcher's ye occurs infrequently and in small clusters of lines, from which he infers that there 'Fletcher has done nothing more than touch up a Shakespearean passage, or insert a passage of his own in a Shakespearean context' (79). But, as Fredson Bowers has persuasively argued, it is more likely that in the disputed scenes, and indeed in those which Hoy agrees with Spedding in attributing to Fletcher, the playwright's partiality for ye, together with his other linguistic preferences, has been to varying degrees obscured by the alterations of the compositors and, more especialy, a scribe: Bowers accepts the reasons advanced by the editors of the Oxford Complete Works for thinking that the manuscript copy for the Folio text of Henry VIII was a scribal transcript, not the playwrights' own papers.(11)
The investigation reported in this note did not extend to analysis of the other Shakespeare-Fletcher collaboration, The Two Noble Kinsmen. Comparisons of phrase lengths across both Shakespeare-Fletcher plays - plays belonging to separate editorial traditions, since Henry VIII has always been included in Shakespeare's canon, whereas The Two Noble Kinsmen has historically been linked more closely with Fletcher's - would entail analysis of variations in the punctuation of several editions. However, when G. R. Proudfoot's Regents edition (London, 1970) of The Two Noble Kinsmen is used and samples are drawn from Shakespeare's and Fletcher's contributions, as these have traditionally been defined, the phrase-length results are similar to those for Henry VIII.(12) Again the sample ascribed to Fletcher employs the higher proportion of phrases of three, four, five, six, and twelve words, and the sample ascribed to Shakespeare the higher proportion of seven, eight, nine, ten, eleven, and thirteen or more words. The only differences are that in this play two-word phrases are proportionally more common in the Shakespeare sample and single words followed by punctuation are more common in the Fletcher sample. The overall percentages of phrases of two, three, four, five, six, and twelve words are 52.5 for 'Shakespeare', 62.9 for 'Fletcher' (compared with 56.4 and 65.0 in Henry VIII).
Table 3: Scenes of Henry VIII ranked according to their percentage
of phrases of 2, 3, 4, 5, 6, or 12 words. Asterisks mark scenes
considered by Hoy to be basically Shakespeare's but with sporadic
insertions and revisions by Fletcher.
Scene % phrases of Spedding
2-6 or 12 words ascription
V.iii 76.0 Fletcher
I.iv 72.5 Fletcher
I.iii 69.1 Fletcher
*IV.i 67.2 Fletcher
*II.ii 65.7 Fletcher
V.ii 65.3 Fletcher
*IV.ii 63.9 Fletcher
III.i 63.6 Fletcher
*III.iib 63.3 Fletcher
V.iv 61.0 Fletcher
*II.i 59.1 Fletcher
V.i 58.9 Shakespeare
III.iia 58.2 Shakespeare
I.ii 57.9 Shakespeare
II.iii 56.6 Shakespeare
II.iv 55.0 Shakespeare
I.i 52.4 Shakespeare
This study tends, therefore, to confirm Spedding's original findings. Some scenes of Henry VIII may well be of mixed authorship, and it remains possible that Shakespeare and Fletcher are not the only playwrights whose hands are in the play. But the evidence of phrase lengths tells against Hoy's view that several scenes assigned by Spedding to Fletcher are substantially Shakespeare's with a few short insertions by his collaborator. In their phrase lengths these disputed scenes are clearly distinguishable from those agreed to be Shakespeare's but indistinguishable from those that Hoy and Spedding concur in attributing to Fletcher alone.
Scene by scene figures for phrase-lengths 1-13+ are as follows: I.i: 20, 21, 34, 39, 28, 28, 26, 19, 31, 14, 7, 3, 22; I.ii: 23, 32, 37, 26, 32, 31, 25, 23, 17, 10, 8, 10, 16; I.iii: 5, 7, 15, 16, 11, 14, 6, 4, 3, 3, 3, 4, 6; I.iv: 10, 25, 29, 29, 24, 21, 10, 12, 8, 1, 5, 4, 4; II.i: 11, 12, 30, 37, 34, 22, 17, 16, 22, 15, 6, 4, 9; II.ii: 16, 23, 28, 35, 21, 25, 10, 14, 9, 9, 3, 4, 10; II.iii: 17, 12, 26, 17, 21, 12, 15, 10, 9, 7, 2, 2, 9; II.iv: 23, 20, 31, 57, 25, 25, 22, 17, 21, 14, 15, 6, 23; III.i: 17, 33, 22, 39, 40, 26, 18, 28, 15, 4, 5, 6, 8; III.iia: 16, 36, 30, 29, 21, 42, 22, 23, 18, 13,7, 3, 16; III.iib: 31, 44, 46, 61, 50, 37, 27, 23, 19, 16, 13, 10, 15; IV.i: 11, 21, 27, 27, 22, 18, 12, 8, 11, 6, 6, 6, 5; IV.ii: 24, 27, 40, 48, 26, 31, 23, 16, 9, 12, 5, 3, 10; V.i: 25, 33, 38, 40, 29, 20, 16, 22, 20, 16, 4, 10; V.ii: 18, 37, 50, 53, 30, 35, 21, 23, 19, 16, 5, 8, 11; V.iii: 4, 5, 17, 15, 8, 10, 3, 2, 6, 1, 2, 2, 0; V.iv: 7, 2, 14, 20, 16, 13, 8, 11, 5, 4, 2, 1, 5.
MacD. P. Jackson University of Auckland
1 James Spedding, 'Who Wrote Shakespeare's Henry VIII?', Gentleman's Magazine, xxxiv (1850), 115-23, 3812; Samuel Hickson, 'A Confirmation of Mr. Spedding's Paper on the Authorship of Henry VIII', N&Q, ii (1850), 198, 401 3, and iii (1851), 33-4; Spedding's and Hickson's contributions were reprinted in Transactions of the New Shakspere Society, i (1874). Independently of Spedding, Ralph Waldo Emerson, in a lecture published in Representative Men (1850) noted and characterized two styles in Henry VIII. Documentation of the authorship debate is provided in Evidence for Authorship, ed. David V. Erdman and Ephim G. Fogel (Ithaca, 1966), 457-78, and Linda McJ. Micheli, 'Henry VIII': An Annotated Bibliography (New York, 1988), ix-xiii, 103-29. G. Wilson Knight's claims were made in The Crown of Life (Oxford, 1947).
2 Cyrus Hoy, 'The Shares of Fletcher and his Collaborators in the Beaumont and Fletcher Canon (VII)', Studies in Bibliography, xv (1962), 71-90. Hoy's meticulous series of articles in Studies in Bibliography, beginning with viii (1956), represented a major breakthrough in research into the authorship of English Renaissance plays.
3 Gary Taylor, 'The Canon and Chronology of Shakespeare's Plays', in Stanley Wells and Gary Taylor with John Jowett and William Montgomery, William Shakespeare: A Textual Companion (Oxford, 1987), 133-4.
4 Robert Boyle, 'Henry VIII: An Investigation into the Origin and Authorship of the Play', Transactions of the New Shakspere Society, viii-x (1886), 443-87; H. Dugdale Sykes, Sidelights on Shakespeare (Stratford-upon-Avon, 1919), 18-47.
5 Marco Mincoff made this point strongly in 'Henry VIII and Fletcher', Shakespeare Quarterly, xii (1961), 239-60.
6 Thomas B. Horton, 'Distinguishing Shakespeare from Fletcher through Function Words', Shakespeare Studies, xxii (1994), 314-35; Jonathan Hope, The Authorship of Shakespeare's Plays (Cambridge, 1994), 67-83; Hope's main data had been presented and analysed in mathematical detail in 'Applied Historical Linguistics: Socio-Historical Linguistic Evidence for the Authorship of Renaissance Plays', Transactions of the Philological Society, lxxxviii (1990), 201-26.
7 Several early studies of sentence length are discussed by C. B. Williams in his Style and Vocabulary: Numerical Studies (London, 1970).
8 Mincoff, 242.
9 The strong metrical association of II.i with Fletcher is demonstrated by, for example, Charles A. Langworthy, 'A Verse-Sentence Analysis of Shakespeare's Plays', PMLA, xlvi (1931), 738-51; Ants Oras, '"Extra Monosyllables" in Henry VIII and the Problem of Authorship', JEGP, lii (1953), 198-213; Robert A. Law, 'The Double Authorship of Henry VIII', Studies in Philology, lvi (1959), 471-86. Regulation rates for auxiliary do in Henry VIII are most fully treated by Hope in 'Applied Historical Linguistics' (1990) and Authorship (1994), as in n. 6 above.
10 In an excellent unpublished paper, 'Stylistic Variation in Henry VIII', David J. Lake shows that when the incidence of several of the more reliable Fletcher and Shakespeare markers is found for twenty-two scenes and sub-scenes of 100-200 lines in Henry VIII, with the totals for Shakespeare markers being subtracted from the totals for Fletcher markers and the result expressed as a single 'coefficient of style', two clearly separable distributions emerge, each approximating to a normal curve; that the same is true for The Two Noble Kinsmen, but that for independent Shakespeare and Fletcher plays the coefficients approximate one normal curve. On Lake's variables, the only doubtfully classified scene of Henry VIII is IV.i, which phrase length figures associate unequivocally with Fletcher.
11 Henry VIII, ed. Fredson Bowers, in The Dramatic Works of the Beaumont and Fletcher Canon, VII (Cambridge, 1989), 3-20.
12 The samples consisted of I.i.25-214, I.ii, V.iii, and V.iv for Shakespeare (848 phrases), and III.vi.l-308 and II.ii.l-153 for Fletcher (859 phrases).