Textual Variants in the Digital Age
By Christopher Cannon
The editing of Chaucer’s Canterbury Tales in the form in which we now read it took many decades of work by a number of different scholars, but there is as yet no readily available edition that takes account of all the different versions in which the Canterbury Tales survives. Some of this is purely pragmatic. There are over 80 surviving manuscripts from before 1500 containing all or some parts of the Tales (55 of these are complete texts or were meant to be). The great Oxford edition of the nineteenth century, by Walter William Skeat, relies mostly on a single manuscript (‘Ellesmere’) with corrections from only six other texts. The edition in which most people have read the Tales in recent decades, the Riverside Chaucer (also printed in the UK by Oxford), also relies on Ellesmere, although it consults many more manuscripts than six to establish its base text. This is also Jill Mann’s practice in the recent Penguin edition of the Tales. But what if someone wanted to edit Chaucer from all the manuscripts, accounting carefully for all the variations? What if a student simply wanted to get some sense of what sorts of variation were possible in a manuscript culture, where every copy of a text was different because every such copy had to be hand-written?
Since 1940 such curiosity could be satisfied in John Manly and Edith Rickert’s seven-volume Text of the Canterbury Tales ‘on the basis of all the known manuscripts’. But this edition has been under a cloud ever since it first appeared because it used all these manuscripts to try to work back to the ‘original’ text of the Tales (the process is called ‘recension’), and accidentally demonstrated in the process that this is not possible. There probably never was such an original, not least because Chaucer never finished the Tales, and even if there had been, there are too many small errors in the extant manuscripts to eliminate all of them. A more practical obstacle for any curiosity about the nature of this variation, however, is the form in which Manly and Rickert had to present the information they assembled. Take just the 10th line of the portrait of the ‘clerk of Oxford’ in the General Prologue of the Tales, for example, in which the narrator tells us his library consisted of
Twenty bookes clad in black or reed
This line can be found on p. 14 of volume 3 of this edition, and if one looks down to the bottom of the page a set variants is displayed in this way:
or] and Ha4 –a –b* (-) cd* (-) Bo2 En3 Fi PS Py Ra3 Tc1
‘Or’ identifies the word for which variants exist; the close bracket marks the start of those variants; ‘and’ is the word that sometimes occurs instead of ‘or’; and the alphanumeric soup that follows ‘and’ consists of the identifiers (‘sigla’) for the manuscripts in which this variant can be found. As it happens, this list is a simplified account of the variants for this line in all of the manuscripts that Manly and Rickert consulted — what it seemed necessary to mention in order to justify the text they printed. The full variation of these variants is printed in the ‘corpus of variants’ that fills the last three volumes of this edition, and the entry for the line I have just quoted (on page 24 of volume 5) is as follows:
bookes] goode b. Ps2 | clad] clothed Ha4 Ii; clodde Ht | blak] whit Cx1 Fi N1 Py S12 Tc1 | or ] and a Bo2 c Cx1 D1 Fi En3 Ha2 Ha4 Ht Ii Ld1 Mg Mm N1 Ps Py ra3 Ry2 Se Tc1 | reed] in r. Dd Ht Ry1
This list shows that some manuscripts say ‘goode bookes’ rather than only ‘bookes’, that some say ‘clodde’ or ‘clothed’ rather than ‘clad’, and that some say ‘white’ rather than ‘black’. None of these differences is hugely consequential (though the last of them is certainly interesting), but the combination of triviality (to meaning) and complexity (of form) will itself be enough to explain why Manly and Rickert place them in separate volumes, or why an editor such as Skeat would narrow the range of manuscripts from which he chose his readings to 7, or why he too would limit the variants he printed to no more than the readings he corrected (few enough to fit in a narrow column at the bottom of each page). In fact, for the line on the clerk’s books, there is nothing at all at the foot of Skeat’s page, although the line printed is slightly different from Manly and Rickert’s:
Twenty bokes, clad in blak or reed
The difference consists of the spelling ‘bokes’ in place of ‘bookes’, and the insertion (this would have been Skeat’s editorial decision) of a comma. More important than the nature or size (or consequence) of this difference, however, is the way that the print technology that has determined how we read Chaucer for so long must always tend toward Skeat’s simplicity. Mann’s Penguin edition, the most user-friendly we now have — and therefore destined to be the most read in future years — does not include variants at all. The Riverside Chaucer includes some variants in its ‘Textual Notes’ but it moves all of these to the back of the book. Complicate things as much as Manly and Rickert did with all the variants, and no one will read your text.
Online editions open out new possibilities for marrying simplicity to completeness. One can imagine a hypertext edition in which all the variants associated with a particular word, phrase, or line, would simply appear as a cursor passed over the text (much as the contents of a footnote will appear in a bubble when the cursor moves across the reference in a text written in Word). The paradox seems stark: only when the words of the text are lifted entirely away from any page can the complexity of the page be fully preserved and disseminated. And yet it is not a paradox if we think of such a hypertext as finally overcoming the limits of print technology. No longer shackled by the limits of mechanical reproduction, the digital age gives us a text that, precisely because it lacks physical form is supple enough to represent the complexity of that form.
Such a hypertext is not yet with us because entering the variants in the marked-up form that would make them available in this way is itself a huge undertaking (digital technology is never more powerful than the information human labor can provide for it). But an online edition such as Oxford’s is already sufficient to the task of making the complex simple in all the ways that a medieval text with many variants requires. If books must separate variants from the text of the Tales in precise proportion to their detail (include many and they must be placed at the back of the book; include all of them and you need several more books), but simply give yourself the virtual page in which an infinite amount of information may un-scroll in one column while the text sits happily, unmoved, in another, and all the variants of a text can accompany every word and phrase and line of that text at all times. Such an edition can also put the complexity of these variants in the hand of everyone — student and scholar alike — who has a computer and an internet connection.
Many libraries own neither Manly-Rickert nor Skeat. And even the copyright library in which I write these remarks, the Cambridge University Library, requires that the volumes of Manly-Rickert be fetched to its ‘West Room’, but keeps Skeat in its Rare Books Room (because it was published before 1900). To bring Manly-Rickert’s variants to Skeat’s text they must be couriered by a member of staff (a reader cannot transport them from West to Rare Books Room himself). Since neither set of volumes circulates I could not bring them home to compare with my own copies of the Riverside and Mann’s Penguin edition. These common editions should have been available on the open shelves (and I could have found them there and brought them to the Rare Books room myself) but, as it happened, on the day I was gathering these volumes together, the Penguin edition of the Tales was checked out.
None of these movements is anything more than tedious, and careful scholarship moves greater mountains of inconvenience very day. And yet these are obstacles that might well defeat the undergraduate or post-graduate who simply wanted to explore what variation might exist in the text of the Tales. And a scholar focused on the variants in the text of the Tales might more easily go beyond those variants (to thoughts about the patterns they display; to theories about the nature or reliability of the edited text itself) if it was easier to consult them. What everybody who reads the Canterbury Tales has lacked up until this point, in other words, is a way of accessing all the richness of the material form in which the Tales survives as a constant and necessary concomitant of a readable text. The representational power of digital technology enriches the works we have long known and loved by just such elaborations.
Christopher Cannon is a Professor of English at New York University and member of the Oxford Scholarly Editions Online editorial board. He teaches Middle English literature at New York University. He took his BA, MA and PhD at Harvard University, and then taught, successively, at UCLA, Oxford, and Cambridge. His PhD dissertation and first book, The Making of Chaucer’s English (1998) analyzed the origins of Chaucer’s vocabulary and style using an extensive database and purpose-built software to demonstrate that Chaucer owed much more to earlier English writers than had been recognized before. His second book, on early Middle English, The Grounds of English Literature (2004), developed these discoveries by means of a new theory of literary form. Most recently he has written a cultural history of Middle English.
Image credit: Image of Chaucer as a pilgrim from Ellesmere Manuscript in the Huntington Library in San Marino, California. The manuscript is an early publishing of the Canterbury Tales. Source: Wikimedia Commons.