How do editors decide how to edit? A number of factors, large and small, come into play, which will determine both the general editorial methodology and some of the local decisions about inclusion and exclusion. Some of these factors involve the work to be edited itself. What kind of work is this? Is it a work of general or specialized interest? Does it survive in one manuscript or many? If many, do its manuscript witnesses suggest it survives in one version or several? If several, which of these versions seem to be by the work's author, which by later redactors—and does this matter?
Given both the work itself and its previous editing history, if any, is the work best represented by producing a single manuscript edition, or is it necessary or desirable to use all the manuscript witnesses? If so, how? Will the edition be a critical edition, which uses evidence from all the manuscript witnesses to create a single text, or a parallel-text edition, which sets different manuscript versions of the same text alongside each other on facing page? Editors must ask these questions not simply in the abstract but with a tough, practical sense of their consequences, for some kinds of edition are much more difficult, uncertain, and above all time-consuming than others.
Furthermore, different publishers favor different editorial policies: TEAMS, for example, publishes only single manuscript editions of works designed to be used by students, whereas EETS generally publishes critical editions, based on single manuscripts but collating all the available textual evidence. Some series, including EETS again and MET, are much concerned with the linguistic information provided by an edition (forms of pronouns and verbs, spellings as a guide to dialect and so on); others are not. In practice, the more idiosyncratic editions, often of major, canonical authors, are also often those published outside a major series, as is true of Chaucer, Langland, Julian of Norwich, and Henryson, among others.
Designed to reproduce on the printed page all the orthographic information provided by a single manuscript of a work (its spellings, punctuation, capitalization, line divisions, marginalia, and rubrication), a diplomatic edition facilitates scholarly study of a work in its manuscript context. Diplomatic editing has now largely been subsumed by Hypertext editing, as well as by manuscript facsimiles, but it can still serve an invaluable function: for example, in the study of holographs (manuscripts written or revised in the hands of their authors), in reconstructing complex copying situations, or in providing routes into very large manuscripts, such as Harley 2253 or the Vernon manuscript. It can also be instructive to consult a diplomatic edition alongside a critical one in order to get a sense of the scale of the editor's interventions, which are usually considerable, especially in areas such as layout and punctuation.
For an example of a diplomatic edition, look at excerpts from E.J. Dobson's edition of the Ancrene Wisse (used with the permission of the Council of the Early English Text Society).
An extension of the diplomatic edition, a parallel-text edition presents multiple versions of a text side-by-side. Again, this has largely been subsumed by Hypertext editing. Parallel-text editions allow scholars to compare versions of a text line-by-line. They can be useful in identifying textual variants in context, or determining which version to use as the base text for a critical edition.
Parallel text editions are also useful for source study. For example, Stephen Barney's prallel-text edition of Boccaccio's Il Filostrato and Chaucer's Troius and Criseyde shows the reader where Chaucer follows his source closely, and where he alters it or interpolates new text.
For an example of a parallel-text edition, look at this early-twentieth-century EETS edition of The Middle-English Harrowing of Hell (used with the permission of the Council of the Early English Text Society).
Any edition that attempts to construct a text of a work using all the available evidence is "critical," whatever its methodology. Critical editions require collation of the different manuscript witnesses, and the construction of a reading text out of the results of that collation. Most critical editions use a base manuscript, whose readings they accept except where there is reason not to do so, but some are more eclectic than others. Critical editions encourage readers to think about the work, more than about its specific manuscript presentation, and may well be more informative on such topics as the work's sources, historical context, form, style, and other literary matters.
Critical editions you may be familiar with include The Riverside Chaucer, edited by Larry Benson, and The Vision of Piers Plowman by William Langland, edited by A.V.C. Schmidt. In general, a critical edition will contain a single (edited) version of the text, along with substantial introductory matter, explanatory, and textual notes.
While any edition that uses digital media can be described as a "hypertext" edition, hypertext editing generally attempts to present larger quantities of information than the printed page is capable of, allowing access to the information by use of pop-ups, text boxes, a choice of reading environment, and so on. Some hypertext editions, such as those of some of Chaucer's poems, contain transcriptions of all manuscript versions and allow for instant comparison and collation of different versions, perhaps with scholarly commentary. See, for example, The Book of the Duchess Hypertext Edition. Others, such as the series of hypertext editions of Piers Plowman, focus on providing up-close access to images of the manuscripts themselves, with accompanying transcriptions, to allow study of such features as spacing, letter forms, and corrections.
Like all editing, hypertext editing has its particular ideology. It is interested in conveying information in a 'non-hierarchic' environment, and empowering readers by giving them access to all the information used by editor. The counter-argument to this philosophy is that hypertext editing deprives readers of the help of a potential expert witness, the person who knows the details of the text better than any other, the editor. Hypertext editing offers readers undigested data, not digested information.