Report from an Editors’ Review of the Modern Language Association’s Committee on Scholarly Editions’ Draft Guidelines for Electronic Scholarly Editions





At the request of the Committee on Scholarly Editions, on Saturday, November 20, 1999, the following editors of electronic scholarly editions and experts in text encoding attended (or sent a representative with a report to) a meeting at the Institute for Advanced Technology in the Humanities at the University of Virginia in Charlottesville, Virginia, in order to review the Committee’s draft guidelines for electronic scholarly editions (from left to right::


John Unsworth, Dept. of English, University of Virginia, Director, Institute for Advanced Technology in the Humanities and current member of the Committee on Scholarly Editions. 


Hoyt Duggan, Dept. of English, University of Virginia and Editor of the Piers Plowman Electronic Edition (University of  Michigan Press)


Bethany Nowviskie, representing Jerome McGann, Dept. of English, University of Virginia and editor of The Rossetti Archive (Institute for Advanced Technology in the Humanities)


Lou Burnard, Oxford University Computing Services, Oxford University and European Editor of the Text Encoding Initiative Document Type Definition.


Daniel Pitti, project manager, Institute for Advanced Technology in the Humanities and editor of the Encoded Archival Description Document Type Definition


Morris Eaves, Dept. of English, University of Rochester, co-editor of the William Blake Archive (Institute for Advanced Technology in the Humanities), and current member of the Committee on Scholarly Editions




Attendees had received in advance the draft guidelines produced by the Committee in 1998 (edited by Charles Faulhaber), and they were asked to evaluate those guidelines in light of their practical experience in electronic scholarly editions, and to report to the Committee on the usefulness, applicability, and sufficiency of these draft guidelines. 




The conclusion of the editors, described in more detail below, was that there should not be separate guidelines for electronic and print scholarly editions.  Instead, the editors recommended that:


  1. The CSE should develop one three-tiered set of guidelines for scholarly editions (regardless of media), of which


    1. the first tier should be a very brief (one- or two-page) document that states in the most durable, most general terms the absolutely necessary characteristics of scholarly editions. This document should be infrequently revised;


    1. the second tier should consist of best practices within specific traditions of editing, and those practices should, in each of their particulars and examples, be coherent with the necessary characteristics.  This document could be revised more frequently, and should include checklists developed within these specific traditions.  It might be appropriate at this tier, also, to specify principles not applicable across traditions (for example, the principle that once published, the electronic edition should never be changed, or should be changed only under certain circumstances, etc.), and


    1. the third tier would be a technical best practices manual, giving examples that are relevant today.  This document would, necessarily, be revised on a regular basis.  This document should include a kind of technical checklist (If the edition claims to be TEI conformant, does it in fact meet the requirements of a TEI-conformant edition?  If it is in SGML, does it parse according to a DTD).


  1. As an appendix, the CSE should develop a bibliography that points to other resources, standards, and articles that provide useful background at any of these levels.


  1. The CSE should vette these top-level necessary characteristics, and develop lower-level document, in conjunction with the ADE and any other scholarly societies that represent significant editorial traditions (at a later stage in the process, in fact, one might want to consult with groups concerned with preservation, cataloging, publishers).  One reason that such a coordinated effort is to be desired is that these scholarly societies might want to have some impact on more demotic standards (such as the Ebook standard) that are now developing in the mass market.




There are two groups of people to whom such guidelines would be of use: those just starting a first editorial project and wanting to know what the rules are, and those who have established editorial practices and want (or for the purpose of peer review, etc., need) some way to determine whether those practices are adequate.  For both groups, the statement of general principles and current best practices would be helpful either as pointers or as reminders.   As for the third tier—the section of the guidelines that lays out technical best practices in different media, as they are understood at the moment—these guidelines would identify best practices for both print and electronic media so that vettors of editions, for example, could have a point of reference for evaluation, but also so that editors could have the same.  In the case of electronic editions, and perhaps only slightly less so in print, there’s a real problem with base knowledge of the vettor as well as the editor.  A checklist, therefore, would be helpful.  Finally, since the world is going to have cheap electronic books and scholarly editions both, an important purpose of the combined guidelines might be to distinguish the two and to identify the qualities of a good electronic edition in the same terms used to identify the qualities of a good print edition, even though in certain respects (e.g., at tier 3) those qualities might be different.



Guidance for further drafts and revisions:


Assuming that the Committee accepts the recommendation of these editors and replaces its current guidelines for print editions and its proposed guidelines for electronic editions with a single, three-tiered document, and assuming that in doing so the Committee favors revision rather than starting from scratch, here are some suggestions for how material in the draft guidelines for electronic editions might be revised and/or carried forward into a new, unified set of guidelines.



Tier 1 (common principles):


This opening section should be the briefest and least frequently revised section of the guidelines.  It should offer no medium-specific advice and no advice that applies only to editions of certain kinds of material or editions in certain editorial traditions.  It should speak only to very general principles: For example, it might advise the editor to specify the relationship between original source material and your representation of it.  If the input is a sample rather than the whole of the available information, the principle of selection should be stated. In general, the idea of this section is to prompt editors to consider and make explicit their methodology.


All scholarly editions probably require, then:


n                          establishing a reliable text and stating the grounds on which reliability is being established (including perhaps, but not necessarily, authenticity)

n                          stating the relationship between edition and source material,

n                          documenting the principle of selection or sampling,

n                          documenting the edition’s method of representation (e.g., “presentation of text” in a print edition, encoding practices and/or stylesheets in an electronic edition)



Thomas Tanselle’s recent essay on “primary documents” is a convenient statement of these ideas (Tanselle, G. Thomas.  “Statement on the Significance of Primary Records,"

Profession 95 (an annual MLA journal), p. 27-28).


The list at the bottom of p.2 (“It must employ non-proprietary encoding standards”) should instead specify characteristics of whatever is considered the “edition itself” (a definition that the committee might wish to offer, or might want instead to require of the editor) rather than the ephemeral delivery mechanisms of the edition.  Granted, this recapitulates the print debate about substantive vs. accidental, etc. etc., but perhaps all we want to say is that because of the cost of producing these things, it is essential that core components of any edition (whether published electronically or in print) must be portable across future systems, must be reusable in other contexts than the one for which they are initially produced.  Indeed, this might imply that all editions should be prepared electronically according to best open-standards practices, whether or not the editors intend to publish them electronically, so that even print editions would be held to the principle of “open standards, open source”.  And finally, the third point in this list hinges on the committee’s decision as to what’s the core component of an edition—or the editor’s decision.  It shouldn’t be part of this tier. 


Also noted: statements in the draft guidelines with respect to specific technologies (character sets, hyperlinks, encoding norms) are already dated, and such references will inevitably date quickly.  It might be better to more generally define these things in the guidelines, and then in examples give dated specific recommendations.  Characteristics described in the general statement designed to endure, dated listing/examples naming specific technologies. (Non-proprietary, open, international standard for the representation of information; NB: it’s not a problem to promote SGML at the more general level, but it is to promote TEI above the level of an example.) General principle: the objective should be maximum portability.  Full stop. 


par. Beginning “The editorial standards” (p.1): “The editor’s basic task” and “the editor’s primary responsibility” are less desirable as statements of essential requirements than that found in the third paragraph of the document, in the sentence beginning “The CSE emphasizes” etc..  The editor’s basic responsibility is to be honest about what he has done.


Section II. Could simply be deleted.  Anything that rises to the level of an actual principle (rather than example or tip) should be moved up to an earlier section. 


Section III should be struck.  It repeats in specific terms general principles that should already have been stated.



Tier 2 (editorial best practices):


This section would be particularly challenging to compose, and in fact would likely need to be commissioned in parts from different scholarly communities (and revised every ten or twenty years).  Commissioning reports from these perspectives would, on the other hand, be the most valuable part of this exercise for the CSE itself. 


The document ultimately produced could and probably should say that there are these disagreements about what constitutes a scholarly edition, here’s where they are, here’s what they’re about.  And in view of that, the purpose of the guidelines is to help editors make explicit their methods and their aims. 


Par. 2, point 3, (p 2), needs to say something like “it should be possible to have information about variant substantive readings, and the way you do it should be appropriate to the materials in question.  One way is to provide an apparatus; another is the apparatus plus all the variant texts; three, all variant texts but no apparatus.  Just the text and a piece of collation software is probably not editorially responsible. It should not be said that “a full-text transcription obviates the need to report variant readings”: never true.  Unless, of course, you’re doing documentary editing.  If you’re editing a document, full-text transcription would be fine, and collation would be meaningless, because the document isn’t a witness to something else.  If you’re editing a text, then it does matter.



Tier 3 (technical best practices):


For electronic editions, open source should be available (for example, publishing only the compiled Dynaweb version of an SGML-encoded edition would flunk unless the source files were also available).  Interesting questions will arise in this section—for example, what is the edition?  Does “the edition” include the functionalities of things like retrieval software, user interface, etc.?  Are look and feel an integral part of the edition? 


The list of “essential requirements” that begins at the end of page 1 is actually a list of requirements that would be essential if you were doing a scholarly edition of printed material; in fact, on that list, the third item is slipping into method rather than describing an essential goal or feature of the edition.  Perhaps what is meant is that the definition of accuracy and the method for arriving at it should be documented.  On the other hand, even for tier 3, the requirement that you point within a document to quotations by hyperlink (par 1., page 1-2) is too specific.  It’s a tip, a hint from Heloise. 


p. 3 Roman I: “since a well-prepared edition” etc. is a sentence fragment: remove the semicolon. In fact, a “well-prepared edition” is likely to be prepared across generations of hardware and software.  Replace “ported” with “refreshed” or “migrated” in that same section.  As for TEI references on this page, and elsewhere, SGML can and should be promoted as a standard, but TEI should not be equated with SGML—it is an example, not a principle.


Section C1-4 should make it clear that things like “Profile Description” and “Revision Description” are examples of the sort of thing that one wants in the documentation of the encoding system, and they are drawn from the TEI header. 


Section D should be rolled up into the earlier description of non-proprietary standards rather than being broken out separately here.  Under section D, a and b are again examples or even tips.


Section E should be rolled up into the earlier discussion of long-term migration…


Another small point: VB3b should say “for printed texts, a history of publication” to parallel 3c.