Transcribo: A Graphical Editor for Transcribing and Annotating Textual Witnesses. Preparing a Historical-Critical Edition of Arthur Schnitzler's Works.

Transcribo is a graphical editor for transcribing and annotating textual witnesses, which is being developed at Trier Center for Digital Humanities and Bergische Universität Wuppertal for the project Arthur Schnitzler: Digitale historisch-kritische Edition (Werke 1905–1931 . While it will be available as an independent stand-alone version, it forms the basis of the project's virtual edition platform which significantly expands its functionality.
Project and Requirements

Funded by the Nordrhein-Westfälische Akademie der Wissenschaften and collaborating with Deutsches Literaturarchiv Marbach, our long-time project aims to create a digital critical edition that makes available both printed works and Schnitzler's estate in an up-to-date and philologically dependable form.
Next to the edited text, the focus will be equally on the materiality and the genetic dimension of the work, which involves various methodological challenges. Along with the complexity and sheer size of the underlying material (> 12000 pages to transcribe) it became obvious very soon that the objectives were not to be attained with existing software solutions (being mostly optimized for the reworking of existing transcripts).
Transcribo was therefore developed from scratch for the efficient creation of large amounts of differentiated, deeply annotated transcriptions, which is reflected in its entire layout.
Transcribo: Features and Interface

The graphical user interface is centered around the digital facsimile, usually the scanned physical witness. This facsimile is duplicated, so that the original can always be viewed unobstructed, while all processing steps take place on the slightly grayed duplicate. This arrangement accommodates the use of multiple monitors and particularly saves a time-consuming jumping back and forth between the image and editor window. Thus, selections in rectangle or polygon shape can be drawn topographically precisely above the graphic and the transcribed text can be entered directly. This also simplifies, if desired, to proceed in a non-linear way and create the transcripts in any order, for instance to treat graphically related, but spatially separated locations in one step.Furthermore, an OCR (with optional image optimization such as contrast and color adjustment) is integrated for the detection of typescripts, providing a raw transcription for the editors to build upon
In doing so, it is essential that any recorded unit can be commented and that philologically and genetically relevant phenomena can be annotated in a uniform way. This is achieved by context menus with a wide and project-specific choice of options. So far, this includes different variants of corrections and changes inserted by author or editors, the labeling of unsafe readings or unidentified graphs, the marking of graphical elements or global attributes of witness and writing instruments. This selection can be extended at will and will be adapted to the textual requirements over the entire course of the project.
Also, transcribed units can be combined into sequences, whether to document comprehensive semantic or genetic correlations such as combined changes or, in case of extensive overrides and insertions, to mark the resulting textual order.
Since each facsimile is a separate unit, no cross-page connections can be documented in the stand-alone version of the software, a shortcoming that is remedied by its connecting to a database system.

Fig. 1: Transcribing and annotating a manuscript
Transcribo and FuD: Extending the Reach

At this point, however, a basic methodological problem arises: The superposition of a semantic, genetic, and material-documentary perspective creates different levels of description which overlap and intersect. Such configurations are very cumbersome to handle by exclusively XML-based data structures or software solutions. This is one of the reasons why the developers refrained from letting the users work directly in an XML environment. Instead, the program is based on an internal data structure, the complexity of which is attenuated by the graphical interface, while the results can be exported at any time to XML/TEI.
Moreover, it allows for a seamless integration into an existing environment, the Research Network and Database System “FuD”[i].
Transferring the content into a centralized database system and associated infrastructure further expands the capabilities. The existing options for annotation can thus be extended to sequences of any size, independently of given material or logical units. Furthermore there are already proven tools for the managing of metadata and marking of related motifs. Also, auxiliary functions such as a graphematic search can be implemented: if an editor is uncertain about a specific transcription, all previously recognized graphical equivalents to a given sequence of letters can be displayed. As outlined in our last year’s contribution, further program modules are currently being developed which allow the visualization and the comparison of writing stages.
References

www.arthur-schnitzler.de (will go online on Nov. 15th, 2013).
For details, see also: dh2013.unl.edu/abstracts/ab-412.html.
Forschungsnetzwerk und Datenbanksystem: fud.uni-trier.de.