Manuscript Annotations as Deltas:First StepsGioele BarabucciCologne Center for eHumanitiesUniversität zu KölnKöln, Germanygioele.barabucci@uni-koeln.deABSTRACTTexts revised on paper are often marked with graphical signsto show what to change and how. The f i eld of genetic textualcriticism uses these signs to study the evolution of literarymanuscripts.This paper describes a technique that is being used to for-malize these graphical changes into deltas. This techniquemakes it possible to identify not only for simple modif i ca-tions like additions or deletions of texts, but also for morecomplex phenomena such as “undone” modif i cations. Theapplication of this technique allows textual scholars to basetheir arguments on quantitative data.1. INTRODUCTIONBefore the widespread adoption of electronic word pro-cessors, authors of literary works such as novels and playsref i ned the drafts of their manuscripts by writing over theold text using a plethora of techniques were used to markwhat should be changed and how: strike-overs, tiny wordsbetween the lines, cryptic symbols, notes on the margin, etc.In the f i eld of genetic textual criticism, scholars study thedrafts of a work. With these drafts they produce the socalled genetic edition: a version of the text accompaniedby notes that help the reader understand the genesis of thetext.During the study of the drafts, the scholar takes note ofwhat has been changed, how and when. This process resem-bles the creation of a delta: parts of the text are chosen asthe base, other parts are used as the target document andsome signs are interpreted as operations (e.g., a “v” for addhere, a line over the text for delete this). In fact, what iscreated is not a single delta, but a series of deltas.In this paper we will provide an overview of the workbeing done at the Cologne Center for eHumanities of theUniversity of Cologne in collaboration with the Center forManuscript Genetics of the University of Antwerp to turnthese graphical changes on the manuscript into collectionsof formal deltas that can be queried and analyzed, provid-Permission to make digital or hard copies of all or part of this work for personal orclassroom use is granted without fee provided that copies are not made or distributedfor prof i t or commercial advantage and that copies bear this notice and the full citationon the f i rst page. Copyrights for components of this work owned by others than theauthor(s) must be honored. Abstracting with credit is permitted. To copy otherwise, orrepublish, to post on servers or to redistribute to lists, requires prior specif i c permissionand/or a fee. Request permissions from permissions@acm.org.DChanges’16, September 13 2016, Vienna, Austria© 2016 Copyright held by the owner/author(s). Publication rights licensed to ACM.ISBN 978-1-4503-4409-8/16/09...$15.00DOI: http://dx.doi.org/10.1145/2993585.2993591ing the textual genetic scholars with quantitative data withwhich answer their research questions. The core of this shortpaper is in Section 3, where a loose def i nition of “undoing achange” is, step by step, turned into runnable code.2. BACKGROUND AND OBJECTIVEThe general aim of this project is to provide scholars withquantif i able data about the changes made by the authors (orby the editors) during the creation of a literary work. Thisdata is needed in order to answer questions such as“was theauthor more hesitant at the end of their career?” or “wasthe author worried about censorship?”. Questions like theseare usually answered by scholars without providing any harddata, just relying on their academic status and on some il-lustrative examples. However, without precise data to backthem, their answers will be hard to justify and able to standthe kind of scientif i c scrutiny that is now getting more com-mon in the Humanities. Recent uses of quantitative data inthe Humanities has led to refusals of long held theories [3] aswell as def i nitive conf i rmations of widespread positions [1].To explain in practical terms the kind of data we wantto extract, we will use the following f i ctional (but plausi-ble) research question: “Has author John Doe undone morechanges during the writing of his f i rst or of his second novel?”.We assume that the manuscripts containing the draftsof the novels have been transcribed in an electronic form.These transcriptions contain the basic text but also all theannotations and revisions marks present in the manuscripts.Now what is left to do is to 1) def i ne what “to undo achange”means, 2) f i nd the occurrences of that phenomenonin the transcriptions and 3) present the results.The next section focuses on point 1 and describes how aloose description is turned into runnable code.3. FROM A LOOSE DEFINITION TO ACONCRETE ENCODINGBecause we want the computer to go through the vastnumber of modif i cations that appears in drafts, we mustencode our requests in a way that is executable by a machine.The following steps illustrate how we turn a loose def i -nition into a set of machine-readable instructions. Detailsabout the steps will be provided in the subsections.1. Fix the operation by def i ning it in a natural language(e.g. “undo is the revocation of an earlier change”);2. Def i ne in a more precise way, but still in a naturallanguage, the concrete variants of this operation (e.g.,
