As part of our push towards Open Access publishing, CERES created the online journal Entangled Religions. It is a peer-reviewed platinum Open Access journal with a focus on religious contact. It started as a simple OJS journal, with a lot of manual editing and typesetting (in Adobe InDesign). In the end, there was just a PDF download for the articles.
For some time now, we wanted to update both the editorial process and the reading experience. The first step towards that goal was the migration to OJS 3 and a new design. However, the editorial process remained the same. We have larger plans for the future, but we did not want to stick with our old process for so long, so we decided to implement some changes that were possible within the current framework. As a result, we’re now introducing a new editorial workflow and typesetting process.
One of our requirements was a single-source publishing model: We want to be able to offer articles in multiple formats (mainly PDF and HTML, but potentially also ePub). The production workflow should be able to create these output formats from a singe source document. Typically, this would be a JATS XML file which is then processed to generate HTML and PDF. However, the Open Source solutions for a JATS based workflow seem still to be in heavy development. The creators of OJS are currently working on an Open Typesetting Stack to implement this in their system. janeway, an alternative publishing system used by the Open Library of Humanities, already supports this, but this would mean abandoning OJS at this point.
Also, articles (typically submitted as Word files) would have to be converted to JATS. And the XML file would have to be checked and proofread, which would require sufficient XML knowledge on part of our editors. There are visual JATS editors in development, e.g. Texture, but they still seem to be unfinished.
After investigating the options for a while, we decided for a different approach. At the core of our workflow, we’re now using pandoc and its flavor of Markdown. Markdown is a lightweight markup language that gained a lot of momentum in the last years. In contrast to other implementations that mainly aim at creating HTML, pandoc defines an abstract document model, in combination with a Markdown serialization, and conversion routines from and to other document formats, including HTML, DOCX, LaTeX and JATS. The abstract model captures most features of scholarly publications, including footnotes and citations, making it a good candidate for an academic publishing workflow. The abstract model has two advantages:
- The simple model makes it easier to harmonize incoming manuscripts and make sure they only use a very limited set of features.
- Being format independent, it allows to generate other output formats than PDF.
Pandoc’s model does have some weak spots, though, most prominently the lack of table and figure references. Also, its output lacks some features we need, for example paragraph numbers. But pandoc is highly customizable: Through command line options, through custom templates, and through filters, which are small scripts that transform the internal representation of a document.
One conceptual issue we had with this approach is that pandoc’s Markdown is not a stable and standardized format like JATS. This makes it less suited for archiving and exchange. But providing a standardized and machine readable version of our articles does not require that this version is also the source format for our processing pipeline. In the future, we plan to add JATS to our list of output formats.
Another issue we encountered when allowing for output versions other than PDF is citation: Typically, articles are cited on the basis of page numbers. The article itself can be unambiguously referenced using journal, volume and DOI (in fact, the latter would be sufficient). But page numbers are the conventional way of addressing passages within the article. This makes HTML a second-class citizen, as one could read the article online, but would have to refer to the PDF when actually citing it. To solve this, we decided to switch to paragraph numbers instead, as used by DHQ and other online journals. To avoid confusion, we decided to leave out page numbers also in the PDF version.
Our new workflow currently looks like this:
- Manuscripts are handed in as Word files (we theoretically can now accept other formats as well, but this hardly ever used).
- They run through regular peer review.
- The accepted version is converted to Markdown. References are added to a Zotero database and handled as native citations in the Markdown document.
- The editor checks the result of the conversion, cleans up the markup, and adds required metadata.
- The document is then converted to PDF (via LaTeX) and HTML.
All articles that we publish from now on are using the new workflow. An example is Nikolay Tsyrempilov’s From the Faith of Lamas to Global Buddhism: The Construction of Buddhist Tradition in Russian Trans-Baikal from the Eighteenth to the Early Twentieth Century. After trying this out for a while and incorporating necessary changes, we plan on also publishing the scripts we use for the pipeline.

A new publishing workflow for Entangled Religions by Frederik Elwert is licensed under a Creative Commons Attribution 4.0 International License.