The Michael Sperberg-McQueen Digital Nachlass ensures the preservation of Sperberg-McQueen’s digital legacy by creating an accessible archive, built to modern standards of digital preservation. All materials will be archived according to international best practices, providing multiple access layers, and made public for research projects as well as for didactic activities to the extent we legally can. The CMSMcQ archive was created in 2025 and is currently operating based on a limited amount of institutional support from the University of Cologne and private donations. We are already working on a funding bid to further develop the project and continue expanding—and as far as possible fulfilling—the potential of the digital archive.

The initial phase of our project focuses on the publication of a bibliography of Sperberg-McQueen’s work, with links to online versions of his published work whenever possible: research articles, webpages, blogs, standards, source code, and running systems. Taking the cue from the modelling discussions found in the CMSMcQ archive, as part of the long tradition of text modelling, we ask what the best formalism for representing complex texts is.

TEI has always used graph formalisms to represent text. Both SGML and TEI represent hierarchical, rooted, ordered, simple, undirected, connected, acyclic graphs, that is, a type of trees. Already in the 1990s, other graphs formalisms were suggested, but had a limited application due to a lack of formal validation systems and also a lack of useful tools. While some sort of cyclic structure can be expressed in XML using XML ID/IDREF links, and this is highly useful for TEI, this is rather an addition to the XML structure than an extension of it, especially when it comes to validation.

Using a less strict graph formalism for text encoding, for instance by including cycles, one can still use significant parts of TEI. We will show examples of encoded documents where the element names and their meaning (as expressed in the text of the TEI Guidelines) are mostly kept. To what extent the meaning changes is one of our research questions we look forward to discussing at the conference. We will also show how the development in user friendly graph encoding and visualisation systems has made the use of less restricted graph formalisms more available, also outside experimental setting. Further to that, we will suggest some consequences for the integration with external ontologies and other knowledge representation systems.

How to develop meaningful translation paths from graph based encoding to XML based TEI? We will show some examples and discuss to what extent the semantics of the encoding can be kept, and to what extent the result will be human readable. We will also discuss how the very conception of a less restricted graph is a scholarly endeavour that uncovers new dimensions of the materials. The flexibility granted by the chosen framework, paired with the development of a rule system for our graphs which enable syntactic graph validation, allow for a higher degree of intellectual creativity and bottom-up modelling compared to other methodologies.