P5: Recommandations pour l'encodage et l'échange de textes électroniques

The primary goal of this revision has been to make available a new and corrected version of the TEI Guidelines which:

• is expressed in XML and conforms to a TEI-conformant XML DTD;
• generates a set of DTD fragments that can be combined together to form either SGML or XML document type definitions;
• corrects blatant errors, typographical mishaps, and other egregious editorial oversights;
• can be processed and maintained using readily available XML tools instead of the special-purpose ad hoc software originally used for TEI P3.

A second major design goal of this revision has been to ensure that the DTD fragments generated would not break existing documents: in other words, that any document conforming to the original TEI P3 SGML DTD would also conform to the new XML version of it. Although full backwards compatibility cannot be guaranteed, we believe our implementation is consistent with that goal.

In most respects, the TEI Guidelines have stood the test of time remarkably well. The present edition makes no substantial attempt to rewrite those few parts of them which have now been rendered obsolete by changes since their first publication, though an indication is given in the text of where such rewriting is now considered necessary. Neither does the present version attempt to address any of the many possible new areas of digital activity in which the TEI approach to standardization may have something to offer. Both these tasks require the existence of an informed and active TEI Council to direct and validate such extension and maintenance work, in response to the changing needs and priorities of the TEI user community.

Two exceptions to the above principles may be cited: firstly, the chapter which originally provided a ‘Gentle Introduction’ to SGML has been completely rewritten to provide a similarly gentle introduction to XML; secondly the chapter on character sets has been completely revised in light of the close connexion between Unicode and XML. The editors gratefully acknowledge the assistance of the ad hoc workgroup chaired by Christian Wittern, which undertook to provide expert advice and correction at very short notice, in the latter task.

The preparation of this new version relied extensively on preliminary work carried out by the former North American editor of the TEI Guidelines, C.M. Sperberg-McQueen. In a TEI working paper written in 1999¹ he sketched out a precise blueprint for the conversion of the TEI from SGML to XML, which we have implemented, with only slight modification.

The Editors would also like to express thanks to the team of volunteers from the TEI community who helped us with the task of proofreading the first draft during the summer of 2001; and to Sebastian Rahtz of Oxford University Computing Services, without whose skill and enthusiasm this new edition would not have been possible.

A substantial proportion of the work of preparing this new edition was funded with the assistance of a grant from the US National Endowment for the Humanities, whose continued support of the TEI has also been crucial to the effort of setting up the TEI Consortium.

Finally, we would like to thank all our colleagues on the interim management board of the TEI Consortium, in particular its Chairman John Unsworth, for their continued support of the TEI's work, and their willingness to devote effort to the difficult task of overseeing its transition to a new organizational infrastructure.

Summary details of the changes made in the present and previous editions are given in their Prefatory Notes, all of which are now reproduced in an Appendix to the present edition: see Prefatory Notes.

To complete the work started in June of this year, the TEI Editors asked for volunteers from the TEI community to proofread the preliminary XML version. 24 volunteers responded to this call during August, and gave invaluable help both by identifying a number of previously un-noticed errors, and by suggesting areas in which more substantial revision should be undertaken in the future. The Editors gratefully acknowledge the assistance of the following individuals during this exercise:

Jimmy Adair, Syd Bauman, Michael Beddow, Steven Bird, Lisa Charlong, Matthew Driscoll, Patrick Durusau, Tomaz Erjavec, Nick Finke, Tim Finney, Julia Flanders, Mike Fraser, Pankaj Kamthan, François Lachance, Terry Langendoen, Anne Mahoney, Gregory Murphy, Daniel Pitti, Rafal Prinke, Laurent Romary, Stewart Russell, Gary Simons, Elisabeth Solopova, Christian Wittern, Martin Wynne.

In addition to error correction, and clear delineation of those sections in which substantial revision is yet to be undertaken for TEI P5, the present draft differs from earlier ones in the following respects:

• Formal Public Identifiers have been introduced as a means of constructing TEI DTDs and an SGML Open Catalog is now included with the standard release;
• Some systematic errors and omissions in the reference section have been removed; the format of this section has been substantially changed, we hope for the better;
• The chapters on obtaining the TEI DTDs and WSDs have been brought up to date; the chapter on modification has been expanded to include a discussion of the TEI Lite customization;
• All examples and cited markup has been checked for XML validity against the published DTDs, and corrected where faulty; examples have been formatted in a (more or less) consistent style.

This is a preliminary version of a revised and fully XML-compliant edition of the TEI Guidelines. Although work on revising and correcting the text of the document is incomplete, by making available this preliminary version we hope to facilitate testing of the XML document type declarations which it describes by as wide a range of TEI users as possible.

The primary goal of this revision is to make available the corrected (May 1999) edition of the Guidelines in a new version which:

• is expressed in XML and itself conforms to a TEI-conformant XML DTD;
• generates a set of XML DTD fragments that can be combined together in the same way as the existing TEI (P3) SGML DTD fragments to form true TEI XML DTD fragments without loss of functionality;
• can be processed and maintained using readily available XML tools instead of the special-purpose ad hoc software originally used for TEI P3.

As noted elsewhere, a number of errors were corrected in the May 1999 edition. A (much) smaller number of errors have also been corrected in this edition, but no new material has been added. We expect the expansion and modification of the Guidelines to become a real possibility in the context of the newly formed TEI Consortium, which has funded the preparation of this present edition.

A major design goal of both this and the previous revision has been to ensure that the DTD fragments generated would not break existing documents: in other words, that any document conforming to the original TEI P3 SGML DTD would also conform to the new XML version of it. Although full backwards compatibility cannot be guaranteed, we believe our implementation is consistent with that goal.

In making this new version, we relied extensively on preliminary work carried out by the outgoing North American editor of the TEI Guidelines, Michael Sperberg-McQueen. In a TEI working paper written in 1999, TEI ED W69, Michael sketched out a precise blueprint for the conversion of the TEI from SGML to XML, which we have implemented, with only slight modification. The current TEI editors wish to express here our admiration for the detailed care put into that paper, without which our task would have been forbiddingly difficult, if not impossible. We would also like to express our thanks to Sebastian Rahtz of Oxford University Computing Services, for his invaluable assistance in preparing this new edition.

We list here in summary form all the changes made in the present edition. Full technical details are provided in documents TEI EDW69 and TEI EDW70, available from the TEI web site.

1. A new keyword TEI.XML has been added. By setting its value to INCLUDE, rather than the default IGNORE, the user can request generation of an XML rather than an SGML DTD;
2. The content models of all elements have been checked, and, where necessary, changed so that they are equally valid as SGML or as XML;
3. The declared value for all attributes has been changed to a form which is equally valid as SGML or as XML;
4. All the examples have been checked for conformance and converted to use XML syntax, where possible. (This process is currently incomplete.)
5. Some errors and duplications in the class membership of elements from the names and dates tagsets have been corrected.

To implement the first of these, we have parameterized the tag omissibility indicators ‘- o’ and ‘- -’ used within element declarations in the DTD. When XML is to be generated, the parameter entities concerned are redeclared with the null string as their value.

The second change was achieved by removing SGML-specific features (ampersand connectors, inclusion and exclusion exceptions, various types of attribute content) from the DTD and revising the syntax of the DTD to conform to XML requirements (specifically in the representation of mixed-content models, and by removing redundant parentheses). In making these changes, we took care to ensure that the resulting content model would continue to accept existing valid documents, though in the nature of things it could not be guaranteed to reject the same set of documents. As further discussed in EDW69 and EDW70, some constraints (exclusion exceptions, for example) which could be carried out by a generic SGML parser using TEI P3 will have to be implemented by a special purpose TEI validator using TEI P4.

Much work remains to be done, firstly in testing the new DTD fragments against as wide a range of TEI materials as possible, secondly in revising the discussion of markup theory and practice within the text to reflect current thinking. A few sections of the current text (the Gentle Introduction to SGML and the discussion of Extended Pointer syntax are two examples) will need substantial rewriting. For the most part, however, we think the Guidelines have stood the test of time well and can be recommended to a new generation of text encoders scarcely born at the time they were first formulated.

Lou Burnard and Steve De Rose (Editors)

Oxford and Providence, May 2001.

No work of the size and complexity of the TEI Guidelines could reasonably be expected to be error-free on publication, nor to remain long uncorrected. It has however taken rather longer than might have been anticipated to complete production of the present corrected reprint of the first edition, for which we present our apologies, both to the many individuals and institutions whose enthusiastic adoption and promotion of the TEI encoding scheme have ensured its continued survival in the rapidly changing world of digital scholarship, and also to the many helpfully critical users whose assiduous uncovering and reporting of our errors have made possible the present revision.

At its first meeting in Bergen, in June 1996, the TEI Technical Review Committee (TRC) approved the setting up of a small working committee to oversee the production of a revised edition of the TEI Guidelines, to include corrections of as many as possible of the `corrigible errors' notified to the editors since publication of the first edition in May 1994, the bulk of which are summarized in a TEI working paper (TEI EDW67, available from the TEI web site).

During the spring of 1997, this TRC Core Subcommittee reviewed nearly 200 comments and proposals which the editors had collected from public debate and discussion over the preceding two years, and provided invaluable technical guidance in disposition of them. We are glad to take this opportunity of expressing our thanks to this subcommittee, whose members were Elli Mylonas, Dominic Dunlop, and David T. Barnard.

The work of making the corrections and regenerating the text proceeded rather fitfully during 1998 and 1999, largely because of increasing demands on the editors' time from their other responsibilities. With the establishment of the new TEI Consortium, it is be hoped that maintenance of the Guidelines will be placed on a more secure footing. Some specific areas in which we anticipate future revisions being carried out are listed below.

These Guidelines are the result of over five years' effort by members of the research and academic community within the framework of an international cooperative project called the Text Encoding Initiative (TEI), established in 1987 under the joint sponsorship of the Association for Computers and the Humanities, the Association for Computational Linguistics, and the Association for Literary and Linguistic Computing.

The impetus for the project came from the humanities computing community, which sought a common encoding scheme for complex textual structures in order to reduce the diversity of existing encoding practices, simplify processing by machine, and encourage the sharing of electronic texts. It soon became apparent that a sufficiently flexible scheme could provide solutions for text encoding problems generally. The scope of the TEI was therefore broadened to meet the varied encoding requirements of any discipline or application. Thus, the TEI became the only systematized attempt to develop a fully general text encoding model and set of encoding conventions based upon it, suitable for processing and analysis of any type of text, in any language, and intended to serve the increasing range of existing (and potential) applications and use.

What is published here is a major milestone in this effort. It provides a single, coherent framework for all kinds of text encoding which is hardware-, software- and application-independent. Within this framework, it specifies encoding conventions for a number of key text types and features. The ongoing work of the TEI is to extend the scheme presented here to cover additional text types and features, as well as to continue to refine its encoding recommendations on the basis of extensive experience with their actual application and use.

We therefore offer these Guidelines to the user community for use in the same spirit of active collaboration and cooperation with which they have so far been developed. The TEI is committed to actively supporting the wide-spread and large-scale use of the Guidelines which, with the publication of this volume, is now for the first time possible. In addition, we anticipate that users of the TEI Guidelines will in some instances adapt and extend them as necessary to suit particular needs; we invite such users to engage in the further development of the Guidelines by working with us as they do so.

Like any standard which is actually used, these Guidelines do not represent a static finished work, but rather one which will evolve over time with the active involvement of its community of users. We invite and encourage the participation of the user community in this process, in order to ensure that the TEI Guidelines become and remain useful in all sorts of work with machine-readable texts.

This document was made possible in part by financial support from the U.S. National Endowment for the Humanities, an independent federal agency; Directorate General XIII of the Commission of the European Communities; the Andrew W. Mellon Foundation; and the Social Science and Humanities Research Council of Canada. Direct and indirect support has also been received from the University of Illinois at Chicago, the Oxford University Computing Services, the University of Arizona, the University of Oslo and Queen's University (Kingston, Ont.), Bellcore (Bell Communications Research), the Istituto di Linguistica Computazionale (C.N.R.) Pisa, the British Academy, and Ohio State University, as well as the employers and host institutions of the members of the TEI working committees and work groups listed in the acknowledgments.

The production of this document has been greatly facilitated by the willingness of many software vendors to provide us with evaluation versions of their products. Most parts of this text have been processed at some time by almost every currently available SGML-aware software system. In particular, we gratefully acknowledge the assistance of the following vendors:

• Berger-Levrault AIS s.a. (for Balise);
• E2S n.v. (for E2S Advanced SGML Editor);
• Electronic Book Technology (for DynaText);
• SEMA Group and Yard Software (for Mark-It and Write-It);
• Software Exoterica (for CheckMark and Xtran);
• SoftQuad, Inc., (for Author/Editor and RulesBuilder);
• Xerox Corporation (for Ventura Publisher).

Details of the software actually used to produce the current document are given in the colophon at the end of the work.

Many people have given of their time, energy, expertise, and support in the creation of this document; it is unfortunately not possible to thank them all adequately. Below are listed those who have served as formal members of the TEI's Work Groups and Working Committees during its six-year history; others not so officially enfranchised also contributed much to the quality of the result.

The editors take this opportunity to acknowledge our debt to those who have patiently endured and corrected our misunderstandings of their work; we hope that they will feel the wait has not been in vain. For any errors and inconsistencies remaining, we must accept responsibility; any virtue in what is here presented, we gladly ascribe to the energies of the keen intellects listed below.

C. M. Sperberg McQueen and Lou Burnard