New Approaches to Tackle Textual Variation

Abstract

Digital humanities (or DH) as an area of research does not draw a clear line between “traditional” and computer-assisted scholarship. In this project, the AnonymClassic team decided, after an initial survey, not to consider the utilization of general-purpose software for specialized DH practices. Instead, we chose to develop dedicated software to support our research. The flexibility of general-purpose software causes the problem of inconsistency, i.e., each scholar produces documents that are not of the same structure and format. Another issue is that much commonly used general-purpose software is commercial; this would force future projects to purchase licenses for this software in order to replicate our workflow, which is an unnecessary financial burden. A further challenge in general-purpose software is that it may require specialized training, which is not always time-efficient over the life of a research project. When it comes to the specific practice of computer-assisted textual scholarship, the use of the TEI/XML standard is quite prominent, and many edition projects center their efforts around producing XML documents. However, this was deemed a poor fit for our purposes, for technical and practical reasons. The main technical issue is that XML is much less effective with larger text-corpora. The TEI standard is also not specific enough for describing Arabic manuscripts, and the procedures for extending the standard are cumbersome. The main practical issue is the steep learning curve for TEI/XML, especially given that scholars in the project are already engaged with complex questions in their areas of study. The team therefore decided to develop a web-based platform to support data entry and edition practices. This choice arose from the fact that everyone is familiar with web applications, and the learning curve would be moderate. Furthermore, all team members have web browsers installed on their computers, so the extra effort of producing software that is compatible with different operating systems will be avoided. Our platform, which is provisionally named the “Kalila Platform,” has a modular architecture, with multiple interfaces to facilitate collaborative development, easy data querying, extensibility, and reusability. The platform is implemented based on what is known as a “microservices architecture” for the backend, and micro-frontends for the user interfaces. Thus far, interfaces for transcription, descriptive data entry, layout analysis, and textual analysis have been implemented, with careful consideration of consistency in the formatting of our data. A platform for our synoptic edition and comparison of manuscript versions (kept separate up to this point) will be integrated, and an interface for text annotation is in planning.