Between bespoke customization and expressive interface: a reflection from the Frankenstein Variorum project

Authors: Beshero-Bondar, Elisa Eileen / Viglianti, Raffaele / Jin, Yuying

Date: Thursday, 7 September 2023, 11:15am to 12:45pm

Location: Main Campus, L 1.202 <campus:note>

Abstract

Perhaps one of the least appreciated and most definitive challenges of working with the TEI Guidelines is their requirement that scholars interpret and adapt them to the bespoke requirements of their text scholarship. It is particularly challenging to explicate overlapping hierarchies with TEI, for example:

• surface and page vs. chapter, stanza, margin notes spanning pages,

• sentence or clause vs. verse line,

• structural divisions that do not align in different versions of a text.

The TEI Guidelines’ Chapter 20 about Non-Hierarchical Structures perhaps best exemplifies the challenges and possibilities of our attempts to reconcile different ideas of hierarchy in the texts that we study and curate (in our case a page-by-page encoding of a manuscript with structurally encoded print editions).¹ Our primary interest in studying these texts may be to investigate how their conflicting hierarchies were expressed at different manifestations—perhaps in alternate manuscripts or in the publication history.

The TEI Guidelines’ Chapter 20 comments on the challenges to processing when we attempt to model multiple hierarchies, especially when they attempt this in inline ways. But the Guidelines largely (and purposefully) leave scholars to their own devices for processing. We determine for ourselves, based on our available resources, what processing we can apply to express our text data modeling. Yet the processing methods we apply can either extend or limit the expressiveness of our scholarly projects in the data visualizations and digital edition interfaces we can design. The processing decisions we make will determine whether and how we can share the meaningful conflict points in multiple structures, versions, or ways of interpreting a text.

Based on our efforts with the Frankenstein Variorum project since 2017, we understand conflicting hierarchies to pose a challenge whose resolution should reflect our research questions and our scholarly document data modeling. In the context of our research questions, we will review our processing and production pipeline, including:

• Document analysis of multiple digital editions of Frankenstein to locate points of alignment and determine normalization strategies,

• XSLT required to process conflicting hierarchies by first flattening all versions and then “raising” edition files with marked variants,

• The recognition of the document data structure and its normalization in the interface processing,

• The design of a static interface that displays and foregrounds variation across conflicting hierarchies.

The pipeline processing that we devised to design the interface has changed from our first proof-of-concept launch of the website to our current effort. The decisions to work with JavaScript frameworks (React with Astro’s CETEIcean plugin) determine how we visualize data from our stand-off TEI “spine,” which stores information about our edition files across their conflicting structures. Our pipeline is certainly “bespoke”—custom-fit to the distinct points of comparison across multiple encoded versions. The step-by-step implementation of this bespoke production pipeline is something we need to revisit, correct, update, reflect on as the basis of the important work of interchange that mediates across distinct data models.² While the interface too often represents a compromise with available tooling,³ a production pipeline designed for interchange may open up possibilities to express the TEI in the interface.

About the authors

Elisa Beshero-Bondar, eeb4@psu.edu, https://orcid.org/0000-0002-1709-2314 (Penn State Behrend)

Raffaele Viglianti, rviglian@umd.edu, https://orcid.org/0000-0003-3094-139X (U. Maryland)

Yuying Jin, yxj5181@psu.edu (Penn State Behrend)

Notes

Contribution Type

Long Paper

Keywords

non-hierarchical structures overlap xml data structure interface production pipeline