Figure 1
Overview of the Bullinger Digital project workflow, illustrating its comprehensive approach to digitising and processing historical correspondence.
Figure 2
A snapshot from the teoirgsed website. The left side displays the scanned printed edition, and the right side shows the OCR output. Note the omission of special characters like ˚u in sch˚ulmeister (EN: principal) (see red ellipses) in the OCRed text.
Figure 3
Visualisation of code-switching on the Bullinger Digital website, illustrated with a letter from Johannes Fabricius Schmid to Heinrich Bullinger (29 November 1563) (Bullinger Digital, 2023). Users can choose to highlight languages in different colours.
Figure 4
An example sentence from the Bullinger letters in its original Early New High German and normalised forms.
Figure 5
An example of a search for a) ENHG wein (EN wine). The search shows results for the words as entered. However, since a normalised text has been indexed, letters containing the words b) win or c) wyn (both valid variants of wein) are also returned.
Table 1
Excerpt of the human vs. the automatic summary of a letter from Heinrich Bullinger to Oswald Myconius, 25. December 1546 (see https://www.bullinger-digital.ch/letter/12771 (Bullinger Digital, 2023)).
| LETTER IN ENG GERMAN AND LATIN | HUMAN SUMMARY (GERMAN) FROM THE BULLINGER EDITION | GPT-4 SUMMARY (ENGLISH) |
|---|---|---|
| Früntlicher, fürgeliebter herr und br˚uder, ich wünschen üch, üwer lieben hußfrowen und dem gantzen hußgesind von gott durch unsern herren ein g˚ut glückhafft jar. Des bedoerffend wir ouch wol, dann die sachen stand gantz spitzig und gfarlich. Gott der allmaechtig woe llesich unser erbarmen. | [1] Gute Wünsche zum neuen Jahr, auch an Myconius’ Gattin [Margret] und an die Haus- angehörigen. Im Hinblick auf die gegenwärtigen Gefahren kann man Glückwünsche brauchen! | [1] Greeting and New Year Wishes: Bullinger extends friendly greetings and wishes for a fortunate new year to Myconius, his wife, and household, noting the particularly perilous and critical state of current affairs. He calls for God’s mercy in these challenging times. |
| Ulm wil nitt vil g˚uts rh˚umbs haben der dappfferkeit. Man mumlet, sy vertrue gend sich gern mitt dem Sanherib; quod tibi dictum volo. Noe rlingen, Gien- gen, Boppffingen, Dinckelspühel, Rotenburg an der Tuber, Hall in Schwaben, alles und noch mee vomm keyser yngenommen! Ich bin gantz angsthafft umb den herrlichen gelerten Brentium. | [2] Die Stadt Ulm wird keine Lorbeeren für Tapferkeit ernten, denn angeblich hat sie sich recht leicht mit dem Sanherib“ [Kaiser Karl V.] ” öhnt! Dies als vertrauliche Mitteilung. Nördlingen, Giengen, Bopfingen, Dinkelsbühl, Rothenburg ob der Tauber und Schwäbisch Hall befinden sich in der Hand des Kaisers. Bullinger macht sich große Sorgen um den gelehrten [Johannes] Brenz. | [2] Military and Political Situation: He comments on the lack of bravery in Ulm and rumors of their alliance with Sanherib (possibly a metaphor for aligning with a powerful adversary). He reports the capture of several towns (Nördlingen, Giengen, Bopfingen, Dinkelsbühl, Rothenburg, and Hall) in Swabia by the Emperor. Bullinger expresses concern for the well-being of a distinguished scholar, Brentius, … |
| … | ||
| Nostri officii fuerit in tantis calamitatibus inhortari populum ad preces assiduas et ad poeni- tentiam. Certo, certo inveniet et nos malum istud! | [4] Im gegenwärtigen Unglück muss das Volk zum Gebet und zur Buße angehalten werden, denn bestimmt wird nun auch die Eidgenossenschaft bedroht! | [4] Religious Duty and Moral Exhortation: In light of these calamities, Bullinger feels it’s their duty to urge the people to constant prayer and repentance, with a certain belief that these troubles will also befall them. |
Figure 6
TrOCR model outputs demonstrating multilingual capabilities with an example of a language switch from Latin to Early New High German.
Figure 7
Comparative analysis of different HTR models, showcasing the efficacy of our multilingual approach in the Bullinger corpus.