Mainly, single source publishing proposes to produce multiple outputs from one, single source. As we already said, with this approach, it’s possible to produce a PDF format for printing, an XML export for a digital platform, or a digital version in HTML format, from one source file. Single source publishing requires a highly flexible source, which usually consists of a set of texts, some metadata and bibliographical data, and some images or other media. Some students or researchers already use the principles of single source publishing; however, it is long and tedious with a classic text processor, which lacks flexibility; it is complex and just a little more flexible with LaTeX, and it is both expensive and difficult with XML. Since the beginning of the 2000s, a few alternatives emerged based on (1) lightweight markup languages, (2) structured bibliographies, and (3) open-source convertors; our case study (Stylo) uses Markdown, BibTex, and Pandoc. Markdown is a markup language used in blogging, README files, Jupyter/R Notebooks, and forums; it is easy to learn and to read for both humans and machines. BibTex is a reference management software that structures references and produces bibliographies according to a given style (such as MLA, APA, Chicago, etc.). Pandoc—known as the Swiss Army knife of publishing (Dominici 2014)—is a powerful alternative for producing academic artifacts from one single source. This library allows the user to convert files from one format to another, allowing for one source to be converted into one of many possible outputs. In this section, we explore what it means and what implications it has for scientific publishing.

Scientific publishing has several requirements since the articles, journals, chapters, and books are complex editorial artifacts. The first group of specifications is the paratextual elements (Genette 2002):

Furthermore, scholarly research and publishing require texts to have a semantic structure, both to enrich its content (especially in XML) and to insure an efficient digital diffusion (structured data is paramount to archiving and retrieving texts in large databases).

The second group of complexities is the academic workflow (Figure 1), or more precisely, the path of the text through the stages of submission, peer reviews, corrections, and publication. The authors, the editors, the publishers, and the distributors all require specific file formats through its circulation and validation, and even in a given format, multiple files are used by different contributors without a way to merge them. In the humanities, where the pressure to publish is significant (“publish or perish” [Auerbach 2006]), the actual process of publication is quite slow and cumbersome; this conflicting injunction is painful to both authors and editors. These paratextual elements are restrictive and should shape both the ideation and implementation of the editing process.

The expression appeared with the first graphical user interfaces: the same file could be viewed on the screen and be produced as a printed artifact. Single source publishing has seen more projects with the development of XML, which allows the developer community to manage only one source for multiple artifacts—including multilingual support. In digital humanities it is especially the XML-TEI that is used for digital publishing and the encoding of textual information. The Text Encoding Initiative (TEI) is a consortium that developed (starting in 1987), promotes, and maintains a set of XML elements, attributes, and entities to represent textual artifacts in digital form. Extensible Markup Language (XML) format is used to store data in a structured, hierarchical way; TEI is thus a scheme that specifies how XML is used in the case of written documents. The widespread adoption of TEI in the description (or representation) of textual artifacts and data means that information produced by one institution or individual can be used by others, and documents can be used and parsed regardless of who encoded them. The extensibility of XML and the ongoing maintaining of TEI mean that any information in a written document can be represented for ease of use, as the XML format is easier to reference, share, and search than physical documents or pictures. XML-TEI makes it possible to produce texts in several formats like HTML or PDF from a unique source and instructions (a software or library such as Pandoc) that manage the outputs.

An earlier example of single source publishing is One Document Does it all (or ODD). Its principle is to write the documentation for an XML (or TEI) schema, the encoding rules, and the customization details all in the same file (Viglianti 2019). More fundamentally, it is about writing code and documentation in a single process, and it is the origin of the word processors: the developers started to write their documentation behind the code—with a typewriter—before the creation of the first word processor, Electric Pencil (Kirschenbaum 2016, 100).

Single source publishing is still an important issue that warrants continuous development, mainly because of the tediousness of the publishing process and of financial constraints: the current process is time consuming and requires a certain expertise.

Since single source publishing is a series of principle rather than a specific tool, it can have many different implementations, the simplest of which being to separate the source, the templates, and the conversion instructions. The source is the content, usually a text with semantic and basic layout information, metadata, and medias. The templates model each final output, for example, the arrangement of information as diverse as the title, the metadata, or the body of the text. Furthermore, the templates explicify the behaviours of certain data according to the output format: titles might not be displayed the same way on a printable PDF and an archival copy. A template would also hold instructions of very precise layout information, such as how to display an alt caption for the figures on the HTML version of an article. The conversion instructions are necessary to assign the appropriate templates depending on the outputs. So, if the source is a semantic document, the template can rearrange the elements (textual and paratextual) from a source into an artifact (HTML, XML, or PDF).

As previously mentioned, textual data is rarely made up of only text, like XML-TEI, which is usually used to represent an actual, physical, written document. This scheme focuses on what the document looks like, but there are other formats, markup languages, and schemes that encode semantic rather than visual information. For instance, conceptual markup structures and labels part of the text with categories such as subtitle, section, citation, and external references. When a presentational markup might only “understand” that a word should be in italics, TEI might have added information in its attribute that expresses why the word is italicized (for instance, it might be a foreign word, a book title, or the author decided to emphasize it).

We made Figure 2 to show a complete workflow with four main operations:

This diagram (Figure 2) shows that templates model the text (differently for each output format) in a new format that includes textual content and metadata. This file is then translated with a “graphic layout” that gives information on the disposition that elements should take on a page, and on the geometry of the page itself. These transformations are not only aesthetic, they represent important editorial decisions about how to render a page in a way that conveys both semantic and graphical information.

This figure contains other information. As we already said, the templates are the modelling of the text according to the final outputs. It is both the distribution of data (text and metadata) and the construction of the artifacts. The “graphic layout” is the visual translation of the templating, like the disposition on the page or the typographical choices. It is not just about aesthetic details, but important decisions about how informations are rendered on a page (printed or HTML).

In choosing an input format, many popular writing tools were considered, but commercial non-standard formats were quickly discarded as the likes of .doc/.docx do not allow the production of richly structured content. On the subject of structuration, XML is a good solution; however, it is complex, not intuitive, and there are not enough customizable tools to write and edit in this format. Despite this fact, a few approaches, based on both single source publishing and XML, like Métopes (Vincent 2020), are effective, although they require some expertise and have a steep learning curve. Lightweight and understandable markup languages, like Markdown or AsciiDoc, are an interesting intermediate solution; for instance, they were used in the Getty experiment with Quire (Fauchié 2020) for museum catalogues.

Raw digital textual data is rarely only made up of words (the .txt format being an exception). There are three different types of information normally found in a file containing text: presentational, procedural, and descriptive markup. The first type is usually hidden to the user; presentational markup is buried in the textual file and gives geometrical instructions to a software (such as Page, Words, or LibreOffice) that are not meant to be seen, nor informative for users. Procedural markup is information on how a text should be read or parsed by a computer and is usually legible by a human (a good example being the markup <em>this text is emphased and more precisely or graphically in italic</em> in HTML). Finally, descriptive markup gives semantic information on the content of a text, for instance, by tagging titles and sections, or by giving attributes to procedural and presentational markup to explain it.

In contrario to input where many decisions can still be made, scientific publishing requires specific output formats: PDF, HTML, and various XML with specific schemas—like JATS (Journal Article Tag Suite, an XML schema for scientific publishing) or TEI. The question is to understand the requirements of output formats to design efficient templates: for example, the XML formats necessitate more metadata than a PDF with specific markups, which can be shared with nothing else than a file name.

As previously mentioned, the emergence of single source publishing has economic origins: if a workflow can be set up around one single source and can produce three or four different formats, it represents a huge savings. It is what we call a multimodal approach: one source but multiple artifacts/formats. With one source, there is no need to manage multiple versions, like the .doc version for the writing and editing step, the .xml version for the digital distribution step, and the .indd version for the printing step (INDD format is for the Adobe InDesign software). In theory, single source publishing is a horizontal workflow since all actors of the publishing chain can participate at the same time—for example, editors (and in some rare cases, authors) on the source, publishers on the metadata, designers on the templates, and editorial engineers on the tool chain. So, editors, publishers, designers, and digital distributors can work simultaneously on the different stages of the publishing chain without having to wait for the previous steps to be completed. This process can support continuous and iterative work; building a publishing chain can be an iterative method.

A horizontal publishing chain makes it harder to legitimize its content: where do we situate the validation, structure, and design of the documents, and how do we deal with the de-structuration of the classic chain? With single source publishing, legitimization is no longer a centralized issue: few different people with different levels of technical skills can modify the content until the final steps of the publishing workflow. However, since they are in control of the templates and publishing software, editors and publishers have final say in the layout without having to burden authors and reviewers with the responsibility to adopt or maintain a particular style. This also opens new possibilities for the publishing of new formats or the republishing of older texts with updated formats or layouts.

Single source publishing is possible only if the editorial teams drop the use of commercial word processing (or at least learn not to depend on them). It is necessary that these teams integrate more tech in their staff, or even better, that publishers need to acquire the technical skills to manage or create publishing chains. A positive (or utopian) situation would be the creation of custom publishing chains by publishers by assembling free or open-source programs/software.

Since the 1980s, the scientific community uses mainly one unique numerical writing paradigm for both writing and editing: different generations of word processors WYSIWYG (What You See Is What You Get) like Microsoft Word, LibreOffice Writer, or Google Docs. Proprietary software (and copycats of proprietary software like LibreOffice Writer for Microsoft Word and Scribus for Adobe InDesign) maintains a confusion between the structure of the contents and the graphic rendering. The academic community needs tools that correspond to its constraints, and that means the conscious encoding of semantic (rather than visual) information. Many communities already work in this paradigm of WYSIWYM (What You See Is What You Mean): Manifold, PubPub, Coko, Métopes, Quire, Stylo, etc. Groups and initiatives such as Métopes (Université de Caen), Editoria (Coko Foundation [Hyde 2021]), Quire (The Getty), or Stylo (Canada Research Chair on Digital Textualities and Huma-Num) are implementations of the single source publishing principles.

In Understanding Media, Marshall McLuhan developed the concept of hybridization or hybridity (McLuhan [1965] 1994, 48–55). According to McLuhan, the hybridization of media produces new media, as media have a generative effect on each other. In the case of single source publishing, we should ask the question of the effect and influence of multiple artifacts on their single source and vice versa. It can be assumed that taking into account an additional output format (whatever it may be) will affect the way the source is structured and managed in the publishing process. It is what we can call a hybridity of the elements, with a fluent content modelling through complex templates.

The role of the templates is to translate a semantic construction into an editorial pattern, in determining how to distribute the information through specific instructions in a meaningful way. Hence, in our case, the hybridity is possible through the work done on templates (how the outputs are shaped).

Written Images was a book subject to a crowdfunding campaign from which several artifacts such as a website, printed books, and printed postcards were produced. From Alessandro Ludovico, the hybridization of forms and formats is a phenomenon that can be observed in the first digital experiments, like electronic literature available online or the first augmented e-books. Electronic literature explored the possibilities of writing and publishing literature in new forms (Rettberg 2014). It is about producing literature that can seemingly only exist on digital supports like computers, since it uses hyperlinks, interconnected texts, or multimedia, such as texts, images, videos, and interactions in the same document. Augmented e-books appeared mostly through technologies such as the iPad around 2010. An augmented or enhanced e-book (James and De Kock 2013) is a special form of digital book that incorporate digital elements and tools that could not be included in a traditional paper publication. Augmented e-books make use of hyperlinks (both within and outside the browsed documents), create popup windows, play videos and sounds, might include interactive or live graphs, and even change according to user interaction (either through individual choices or according to a community’s practice).

In Post-Digital Print, Alessandro Ludovico analyzes several examples of hybrid publishing initiatives, where few artifacts derived from the same content complement each other. Electronic versions of books, journals, magazines, or articles came in to complete already existing forms: printed and digital artifacts become hybrid. Even if the source is unique, it is possible to conceive different scenarios depending on the output format; for example, a specific block of text can be displayed differently depending on the artifact. This can be considered as an anticipation or as an investment in order not to waste time on adjustments that would concern artifacts that would no longer have any link between them. Paper publishers already experiment with different versions for a given printed book with the same content inside: from the cheap and affordable paperback to the luxurious, embossed, limited edition hardcovers. It is also not uncommon to find that the digital version lacks some content that would not be well rendered in the EPUB format, or better quality/more pictures in an electronic version where colour printing is too expensive. Hence, single source publishing implements Ludovico’s concept of hybridization: rather than editing several source versions to get the corresponding artifacts, it is possible to create a unique source and use templates as scenarios for generating different artifacts in both their form and content. Hybridization is more coherent and powerful when all of the content of a given project is in the same place, generating different forms and version from the same source.

According to Marcello Vitali-Rosati, editorialization is an evolution of the concept of edition in a contemporary context by taking into account the digital. Editorialization expresses the idea that actions like writing or reading are shaped by the tools and support (platform) we use. The adaptation of publishing chains to produce different artifacts is one such way of understanding and constructing this space. Single source publishing is one expression of editorialization, where all work converges in a horizontal structure in order to produce textual artifacts. The modelling of our space depends on the way we design and build these processes: methods, tools, free/open-source software, technical approaches, etc. Single source publishing implies a remodelling of text within many scenarios, a task that mirrors the shaping of digital and real spaces.