"New Territories". Text Encoding Initiative Conference and Members' Meeting 2025. September 16–20, 2025. Kraków, Poland. Book of Abstracts

Text Encoding Initiative Conference and Members’ Meeting 2025 Kraków, Poland September 16–20, 2025 Book of Abstracts Wydział Filologiczny, Uniwersytet Jagielloński al. Adama Mickiewicza 9, Kraków Table of Contents 3 “New Territories”. Text Encoding Initiative Conference and Members’ Meeting 2025. September 16–20, 2025. Kraków, Poland. Book of Abstracts Edited by: Joanna Hałaczkiewicz Organized by: Jagiellonian Centre for Digital Humanities al. Mickiewicza 11, 31-120 Kraków [email protected] tei2025.confer.uj.edu.pl ISBN: 978-83-977695-0-2 DOI: 10.5281/zenodo.17312233 Layout and cover design: Patrycja Wojkowska Typesetting: Joanna Hałaczkiewicz Proofreading: Kinga Moskal, Wiktoria Szelest Disclaimer The content of each abstract is the sole responsibility of its author(s). The editors and organizers are not responsible for the views expressed in the abstracts. License This publication is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0) Table of Contents 7 Introduction 9 From the editors 11 Keynotes 13 A Case for Best Practices in TEI Encoding of Newspapers. The Darmstädter Tagblatt as an example 15 A new attribute class for annotating syntactic dependency relations 20 A TEI-based Layout Annotation System for a Deeper Automatic Encoding of Documents 28 An XML-based edition publication model 31 Between Language and Music. A Case Study on Encoding of Textbooks on Japanese Traditional Music 33 Beyond Rule-Based Processing. LLM-Assisted TEI Encoding of Editorial Interventions in Historical Correspondence 37 Can we make an AI respect TEI XML? An Experiment with a Small-Scale Explainable AI Model 40 Converting data from Czech digital libraries to TEI 44 Differentiating ODDs 46 Dispatches from the TEI’s GIS Working Group 48 Editing Multilingual Grand Vizierial Correspondence in TEI. The GraViz project 50 Encoding a Work of Interactive Fiction. TEI, Flash, and the Afterlives of ApertureScience.com 53 Encoding Ambiguity. A TEI Approach to Argentinean Reading Theory in an 80s Literary Magazine 55 Encoding Argentinian Eighteenth and NineteenthCentury Drama with TEI 58 Encoding Complexity. TEI Modeling in the “Forschungsportal BACH” Project 60 Encoding Hyperfiction. Preliminary Considerations for a Digital Edition of Susanne Berkenheger’s “Hilfe! Ein Hypertext aus vier Kehlen” (1998) Proofreading of this publication has been supported by a grant from the Priority Research Area (PRA Heritage) under the Strategic Programme Excellence Initiative at Jagiellonian University. 4 Proofreading of this publication has been supported by a grant from the Priority Research Area Proofreading of this publication has been supported by a grant from the Priority Research Area 5 63 Encoding language diversity in TEI. A description of regional and non-standard languages in multilingual diachronic corpora 68 Encoding Ligature Glyphs in Arabic Manuscripts Using TEI-XML 70 Encoding Multilingual and Multiscript Sources in TEI. A historical Spanish-Chinese dictionary 76 Enriching textual data from digital libraries 79 Exploring correspSearch v3 as a Service for Minimal Editions of Correspondence 84 Eye Rhyme in the Digital Victorian Periodical Poetry Project 87 FAIR research data from Goethe’s inbox: The first 2,400 incoming letters as TEI-XML full texts 93 Foregrounding Text Analysis in Digital Scholarly Editing: Who Are We Encoding For? 98 From dialect features to structured data. Modelling spoken Arabic varieties in the WIBARAB project 102 From Practice to Framework. Model Digital Scholarly Editions in the Jagiellonian Digital Platform 106 Functors and format conversion 109 Generating TEI Documents Through a Game of Dice 111 Generative AI for XML-TEI Encoding 114 Graph-Based Digital Editions and TEI. Towards Interoperable and Assertive Scholarly Editing 119 Haute couture (for the masses) 121 Introducing UDraCor. The Ukrainian Drama Corpus in TEI 124 Ioannes Dantiscus’ Itinerary Online 126 Jinks / TEI Publisher 10 128 JinnTap. A browser-based TEI-XML editor 130 LEGOstyle. Building modular, flexible Editions with TEI 134 Linking Your *-ographies. Developing project-specific TEI Authority File Lookups for LEAF-Writer 136 Managing lexical complexity. ODD chaining for Arabic dialect dictionaries 140 Multimodality and Minimal Publishing. TEI, MEI, and more in 19th-Century Music Treatises 145 Navigating and Processing Data from the TEI with XPath and XSLT 147 New features of Scholarly XML, an Open Source Visual Studio Code Extension for TEI encoding 150 New Territories for (Very) Old Language: DiaCorPolL. Automated Compilation of the Diachronic Corpus and Dictionary of Latin in Poland 154 New Territories for Digital Lexicography. Dictionary of Polish Dialects meets TEI 157 ODD-API. A REST Interface for Programmatic Use of ODD Schema Definitions 161 Parallel Editing in TEI. The Case of Regestra 1561 163 Processable and Computable Media in TEI. Usecases and Strategies 167 Reconstructing the Experience of a Historian from the Meiji (明治) Period. Attempts to Encode Kengo Murakawa (村川堅固)’s Diary in TEI 171 Save it on a Shoe String. Experiences with Project Endings 173 Standardisation and innovation. The role of TEI in improving innovative potential in digital scholarly editions of correspondences 175 Starting from Showing, Moving toward Sharing. Broadening TEI’s Reach 177 Streamlining TEI Workflows. Collaborative Editing with NormaTEI 183 Structural Markup of the Database of Historical Polish Lexicons. The Case of Forytarz języka polskiego by Jan Ernesti and Nowy dykcjonarz, to jest Mownik polskoniemiecko-francuski by Michał Abraham Troc 186 Structures and Tools for the Representation and Visualization of Knowledge Contained in Electronic Editions/Textual Resources 188 TEI Processing Model 190 The CMIFerator. A generalised pipeline for contributing to correspSearch Introduction 7 193 The TEI Critical Apparatus at 30-something. Where do we go from here? 195 Towards a model of transcultural encoding of ancient epigraphic sources 198 Towards an Encoding Practice for Multilingual Textual Variation 201 Voci dall’Inferno. A TEI-Based Digital Archive for finding Dante in Concentration Camp Testimonies 205 What do we Need to Make Documents from Texts? 209 Who Knows What a Revision Is? Towards a Shared Vocabulary of Textual Variation 211 Program Committee 212 Local Organizers Introduction It was my great pleasure to see the 25th Annual Meeting of the Text Encoding Initiative—my beloved scholarly community—take place in Kraków, my hometown and favorite city. On paper, TEI is a technical standard and a non-profit organization. In reality, “we are the TEI” (as the title of our closing session with Julia Flanders states). It’s a network connected not so much by formal membership but rather by curiosity and a willingness to explore the fertile crescent where diverse intellectual cultures meet. This diversity is clearly reflected in this year’s program, with subjects ranging from frameworks and interfaces to linguistics; from music and literature to the conceptual and technical underpinnings of TEI itself; from nuanced representations of irony to accountable AI. Is throwing such a wide spectrum of scholarly interests together a feature or a bug? Quoting the opening paragraph of the TEI website, “we develop the Guidelines, which provide the infrastructure for developing machine-actionable cultural heritage texts.” Therefore, this richness is exactly what TEI was designed for: encoding the world, not just the words; providing a platform for hybrid voices and approaches—historical sources and new interpretations, standardization and heterogeneity— all at the same time. This is also what I return to every year at the TEI meeting: great, generous minds and ODDs (forgive the pun), always in favor of lively debate, finding intellectual inspiration, and sowing seeds for fruitful collaborations in the future. This book of abstracts can only be an echo of the event; nevertheless, it gives a taste of some of the good things we enjoyed. Magdalena Turska Chair of the Programme Committee From the editors 9 From the editors Dear Readers, as you flip through this Book of Abstracts, you will encounter voices from across the TEI community exploring textual scholarship, markup, digital editions, and the infrastructures that support our shared work. The themeNew Territories gestures toward both geographic and intellectual horizons, encouraging us to discover together how TEI may unfold in new places, contexts, practices, and collaborations. Kraków, with its rich history and lively academic community, was a fitting place for our meeting. As Poland’s oldest university, the Jagiellonian offers both a chance to reflect on tradition and an opportunity to engage with digital scholarship in fresh ways. This volume is more than a program digest: it is a map of current experiments, emerging standards, and conversations that point to where TEI might go next. You will find abstracts addressing interoperability, machine learning for encoded texts, multimodal markup, regional efforts in underrepresented contexts, sustainable infrastructure, and more — reflections of the community’s shared commitments. We are enormously grateful to the authors, reviewers, organizers, and volunteers who shaped this program; and above all to you — the participants — for bringing your energy, insights, and openness to collaboration. The conference was a true space of discovery, dialogue, and new connections — a moment when territory became not a limit but a horizon. With best wishes, Magdalena Komorowska on behalf of the TEI 2025 Local Organizing Committee Keynotes Perspectives on New Territories in Digital Editing Jadwiga Kita-Huber Jagiellonian University An Introduction to the European Research Council and why it matters for the DH/TEI community Sebastian Winkler European Research Council We are the TEI Julia Flanders Northeastern University A Case for Best Practices in TEI Encoding of Newspapers 13 POSTER A Case for Best Practices in TEI Encoding of Newspapers The Darmstädter Tagblatt as an example Kevin Kuck ULB Darmstadt, Germany Keywords TEI encoding, newspapers and periodicals, best practices, AI, digitization Abstract This poster addresses the reuse of periodicals and newspapers and proposes a best practice for TEI encoding in digitization projects, focusing on the Darmstädter Tagblatt. As part of the newspaper working group of DHd – Association for Digital Humanities in the German Speaking Areas, we recognize the growing need for standardized TEI encoding to facilitate data reuse across various newspaper projects. Leveraging insights from a recently initiated series of workshops on the reuse of newspaper data, hosted in Darmstadt in 2024 and in Vienna in May 2025, the objective is to explore the potential of TEI in enhancing access to historical newspaper data while fostering collaboration among researchers. As the project advances, a range of ideas is taking shape. One such idea is the creation of a universal TEI header that is suitable for both newspaper and periodical editions. An ongoing digitization project at A new attribute class for annotating syntactic… 15 LONG PAPER ULB Darmstadt, which is funded by DFG (German Research Foundation), will serve as the scientific basis. The poster will provide an overview of the project “a Darmstadt Newspaper in Three Centuries - Digitization of the Darmstädter Tagblatt, 1740 – 1986, (Thomas Stäcker, Marcus Müller, Dario Kampkaspar, et. al.)”, highlighting its significance, which also extents to AI/ML research. At TU Darmstadt, an AI assistant is being developed using one of Germany’s longest-running periodicals as a basis. It represents a heterogeneous data set and provides an excellent case study for establishing a best practice. It will also address the challenges and opportunities in TEI encoding of newspapers, proposing general best practices. Engaging with the international TEI community, we seek to foster discussions on standardization, collaborative markup, and the revitalization of the SIG “Newspapers and Periodicals.” By establishing standardized TEI encoding practices for newspapers, we aim to facilitate collaboration, enhance access to resources, and advance research in digital humanities. A new attribute class for annotating syntactic dependency relations Piotr Banski (1); Andreas Nolda (2); Harald Lüngen(1) (1) IDS Mannheim, Germany; (2) Berlin-Brandenburg Academy of Sciences and Humanities (BBAW), Germany Keywords dependency annotation, lightweight grammatical annotation, Universal Dependencies Abstract Overview Language corpora, used as data for linguistic research and machine learning, are traditionally annotated with lemmas, part-ofspeech tags, and, possibly, morphosyntactic categories. As a lightweight, inline, representation of such (mostly automatic) annotations, the TEI Guidelines provide the att.linguistic class with @lemma, @lemmaRef, @pos, @msd, and @join attributes (cf. Bański et al., 2018). Increasingly, corpora are also annotated with syntactic parses in terms of dependency relations between tokens, as a basis for syntactic queries or inferences. A de facto standard for that, supported by a wide array of tools, is the Universal Dependencies framework (UD; cf. Marneffe et al., 2021) with its CoNLL-U format. The present contribution outlines a potential extension of the TEI Guidelines for annotations of syntactic dependency relations, in the form of a new attribute class 16 A new attribute class for annotating syntactic… A new attribute class for annotating syntactic… 17 called att.linguistic.dependency, which extends att.linguistic with the attributes @head and @deprel and several conventions. Example As a rule, syntactic dependency relations between tokens can be modelled by annotating each token with a reference to its syntactic head (if any) and with a label that states the linguistic type of the dependency. Such an approach also underlies the plain-text, column-based CoNLL-U format, where each token is annotated, inter alia, with a numerical index in the ID column, a reference to the index of its head in the HEAD column, and a label of the type of the dependency relation in the DEPREL column. The proposed addition makes it possible to represent this kind of information in TEI XML, as in the following example. <s> <w n=”1” head=”2” deprel=”nsubj” pos=”PRON” lemma=”she”>She</w> <w n=”2” head=”0” deprel=”root” pos=”VERB” lemma=”buy”>buys</w> <w n=”3” head=”2” deprel=”obj” pos=”NOUN” lemma=”book” join=”right”>books</w> <pc n=”4” head=”2” deprel=”punct” pos=”PUNCT” lemma=”.”>.</pc> </s> Proposed TEI encoding of syntactic dependency relations in the sentence She buys books. Note that the <s> element is not the only container possible. The tokens are marked up with <w> tags for words and <pc> tags for punctuation characters, the global @n attribute is used for the numerical token index, the proposed @head attribute holds the numerical index of its head, and the proposed @deprel attribute labels the type of the dependency relation. The root node (typically, but not necessarily, a verb) may either be marked up implicitly as in traditional dependency syntax (cf. the overview in Heringer, 1993) by omitting @head and @deprel attributes, or explicitly as in CoNLL-U with a @head value “0” and a @deprel value “root”. In typical use cases, also lemmas, parts-of-speech tags, and, possibly, morphosyntactic categories are annotated using the @lemma, @pos, and @msd attributes, corresponding to the CoNLL-U columns LEMMA, UPOS, and FEATS, respectively; the @join attribute signals that a token is directly adjacent to another. The proposed lightweight annotation scheme is also capable of representing more complex syntactic annotations, involving concurrent part-of-speech tagging (CoNLL-U column XPOS) or enhanced syntactic dependencies (CoNLL-U column DEPS) as they are used in UD for coordination ellipsis and related syntactic phenomena. The ‘vanilla’ as well as advanced uses of the proposed class will be discussed and illustrated during the presentation. Summary The proposed attribute class att.linguistic.dependency with the attributes @head and @deprel extends the lightweight means provided by the TEI Guidelines for encoding linguistic corpus annotations in TEI XML from the level of token-related properties (defined mostly by att.linguistic) to the level of inter-token relationships. It may be used to extend existing collections of TEI-encoded texts by syntactic dependency annotations, as well as in dedicated NLP pipelines for new annotated corpora. Being lightweight by design, it does not pretend to be a replacement for full-fledged structural and grammatical descriptions enabled by specialised TEI components, such as those underlying standoff representations (Guidelines, ch. 17.10), graph representations (Guidelines, ch. 20.1), tree-based constituent structure representations (Guidelines, ch. 20.3 and the att.segLike class with Between Language and Music… 31 with interaction or searching. The discussion will present a number of design issues for an EPM. References Boot, P. (2024). The TEI Processing Model: Introduction, limitations and potential extensions. In Declarative Amsterdam 2024. https://doi.org/10.1075/da.2024.boot. tei-processing-model. Fechner, M. A (2018). Standardized Interface for Digital Scholarly Editions. DHd 2018. https://edoc.bbaw.de/frontdoor/index/ index/docId/3327. Ide, N. M., & Sperberg-McQueen, C. M. (1995). The TEI: History, goals, and future.Computers and the Humanities,29, 5-15. Turska, M., Cummings, J., & Rahtz, S. (2016). Challenging the myth of presentation in digital editions.Journal of the Text Encoding Initiative, (9). Between Language and Music A Case Study on Encoding of Textbooks on Japanese Traditional Music Shintaro Seki RIKEN, Japan Keywords traditional Japanese music, musical notation, text and character Abstract The Text Encoding Initiative (TEI) has developed and maintained guidelines for the digital encoding of literary and linguistic texts. However, some materials use characters not as symbols for constructing literary or linguistic works, but for other purposes. This poster examines musical scores used in traditional Japanese music as a representative example of such materials and explores possibilities for their digital encoding. Musical scores are documents designed to record music and convey essential information for performance through various types of symbols. In Western musical traditions, a specialized notation system has been developed to visually represent musical structures. In contrast, in Japanese musical culture, characters used in everyday language are often repurposed for musical notation. These are frequently syllables—known as kuchishōga in Japanese—sung to aid in memorizing melodies but POSTER Beyond Rule-Based Processing… 33 generally lacking inherent linguistic meaning. Furthermore, musical scores may also include lyrics and explanations of instrumental techniques. As a result, musical symbols and linguistically meaningful text coexist within the same character system. This poster presents a case study of a music textbook that incorporates both musical symbols and linguistic text, focusing on how its content can be encoded using XML. It explores encoding strategies that accommodate the simultaneous use of symbols from the same character system for multiple functions within a single document, with particular attention to the coexistence of musical notation, semantic content, and accompanying text such as lyrics and performance notes. Additionally, the poster discusses current challenges in linking other notation systems, such as five-line staff notation transcribed using the Music Encoding Initiative (MEI) or Music XML, and linking text materials with time-series media such as audio and video recordings. Beyond Rule-Based Processing LLM-Assisted TEI Encoding of Editorial Interventions in Historical Correspondence Sabrina Strutz, Martina Scholger, Georg Vogeler University of Graz, Austria Keywords Large Language Models, TEI Encoding, Digital Correspondence Editions, Automated Annotation, Editorial Interventions Abstract Editions originating from print or created within the print tradition require specific post-processing to fully realize their potential in the digital realm. This requires critical reinterpretation of editorial conventions which are often implicitly used by the edi - tors and not fully explained in the editorial. The recording of editorial interventions is a central task in the digital edition pipeline for the creation of a critical text (Pollin et al., 2025). While XML-TEI offers a flexible framework for encoding such editorial interventions (TEI Consortium, 2025), actual practice—particularly in printed editions—often suffers from inconsistency and ambigu - ity. This paper examines the potential of transformer-based language models to resolve editorial ambiguities through contextual analysis. One illustrative case is the use of square brackets LONG PAPER 34 Beyond Rule-Based Processing… Beyond Rule-Based Processing… 35 representing different editorial actions as in H[ochwohlgeboren] (abbreviation), emporschwin[gen]des (addition), welches[sic] (indicating errors in the original), [Lücke] (omission) or [?] (illegible content). This kind of ambiguous notations in print editions— including various kinds of brackets or verbal descriptions of textual phenomena in footnotes—pose even more challenges (Beyer 2018, 32; Heckmann, 2013; Leeb, 2013; Schwäbische Forschungsgesellschaft, 2001). The different functions of the editorial notation might be clear to a human reader without explanation, but challenge rule-based processing. LLMs’ ability to transform unstructured text into semi-structured formats and their potential to streamline complex tasks make them promising for Digital Humanities practices. Beyond encouraging results in annotation tasks using generative AI (Ding et al., 2023; Gilardi, Alizadeh and Kubli, 2023; Kuzman, Mozetič and Ljubešić, 2023), recent experiments suggest these technologies could effectively support the labor-intensive annotation phase in digital edition projects (Czmiel et al., 2024; Pollin, 2024; Scholger, Strutz and Pollin 2024). We present a case study using the correspondence of Austrian orientalist and diplomat Joseph von Hammer-Purgstall (1774-1856), where over 3,500 letters from a print edition are being transformed into a digital edition reflecting current DH standards. Unfortunately, clearly defined editorial guidelines are absent or only partially recoverable. While structural elements of the letters can be quite reliably extracted from Word templates via TEIGarage and pattern matching in XSLT into TEI, the ambiguous and inconsistent use of square brackets poses a significant challenge requiring semantic understanding beyond deterministic methods. Our approach extracts segments in square brackets including surrounding context, applies tailored prompts (zero-shot, few-shot) to proprietary and open-source LLMs (GPT-4o, Claude 3.7 Sonnet, OLMo 2), and converts results into TEI en - coding (e.g., <choice><abbr>H</abbr><expan>H<ex>ochwohlgeboren</ex></expan></choice>). A comparative evaluation is based on a manually annotated gold subset. Beyond technical implementation, we address crucial methodological questions: How can we prevent transformer-based models from “over-modernizing” (Scholger et al., 2025) historical content? How can we ensure encoding consistency across large datasets? This research contributes to emerging TEI practices by exploring how generative AI can complement human editorial work while maintaining scholarly standards—recognizing that as LLMs integrate into digital scholarship, their capabilities must be assessed not only for what they can resolve but also for what they might obscure. References Beyer, B. (2018) Praktische Tipps für die Edition landesgeschichtlicher Quellen. Materialien der Historischen Kommission für Westfalen, Bd. 15. https://www.lwl.org/hiko-download/HiKoMaterialien_015_(2018-03).pdf. Czmiel, A., Dumont, S., Fischer, F., Pollin, C., Sahle, P., Schaßan, T., Scholger, M., Vogeler, G., Roeder, T., Fritze, C., and HennyKrahmer, U. (2024, February 21) ‘Generative KI, LLMs und GPT bei digitalen Editionen’. DHd 2024 Quo Vadis DH (DHd2024), Passau, Deutschland. https://doi.org/10.5281/ zenodo.10698210. Ding, B., Qin, C., Liu, L., Chia, Y. K., Joty, S., Li, B., and Bing, L. (2023, June) ‘Is GPT-3 a good data annotator?’ arXiv. https://doi. org/10.48550/arXiv.2212.10450. Gilardi, F., Alizadeh, M., and Kubli, M. (2023, July) ‘ChatGPT outperforms crowd workers for text-annotation tasks’, Proceedings of the National Academy of Sciences, 120. https://doi. org/10.1073/pnas.2305016120. 36 Beyond Rule-Based Processing… Can we make an AI respect TEI XML? 37 LONG PAPER Heckmann, D. (2013) ‘Leitfaden Zur Edition Deutschsprachiger Quellen (13.-16. Jahrhundert)’. Preußenland 3:7–13. Höflechner, W., Wagner, A., Koitz-Arko, G., and Kowatsch, S. (2021) Joseph von Hammer-Purgstall 1774-1856 : ein altösterreichisches Gelehrtenleben : Eine Annäherung. (Vol. 93, Forschungen zur geschichtlichen Landeskunde Steiermark). Graz: ADEVA. Kuzman, T., Mozetič, I., and Ljubešić, N. (2023, March) ‘ChatGPT: Beginning of an End of Manual Linguistic Data Annotation? Use Case of Automatic Genre Identification’. Tech. rep., arXiv. https://doi.org/10.48550/arXiv.2303.03953. Leeb, J. (ed.). (2013) Der Reichstag zu Regensburg 1556/57. 2 Bde. München: Oldenbourg Verlag. McGillivray, B., Poibeau, T., and Fabo, P. R. (2020) Digital humanities and natural language processing: “Je t’aime... moi non plus.” Digital Humanities Quarterly, 14(2). https://www.digitalhumanities.org/dhq/vol/14/2/000454/000454.html. Pollin, C. (2024) ‘Workshopreihe “Angewandte Generative KI in den (digitalen) Geisteswissenschaften” (v1.1.0)’. https://doi. org/10.5281/zenodo.10647754. Pollin, C., Fischer, F., Sahle, P., Scholger, M., and Vogeler, G. (2025, forthcoming) ‘When it was 2024 - Generative AI in the field of digital scholarly editions’, Zeitschrift für digitale Geisteswissenschaften. Scholger, M., Strutz, S., and Pollin, C. (2024) ‘Empowering Text Encoding with Large Language Models: Benefits and Challenges’. Zenodo. https://doi.org/10.5281/ zenodo.13969082. Scholger, M., Cugliana, E., Fischer, F., Pollin, C., Sahle, P., and Vogeler, G. (2025, April 28) ‘Bias und Evaluation - Beiträge zur Kritik der digitalen Edition mit generativer KI’. Zenodo, https:// doi.org/10.5281/zenodo.15299432. TEI Consortium, eds. (2025) ‘4.3.2 Simple Editorial Changes’. TEI P5: Guidelines for Electronic Text Encoding and Interchange. 4.9.0. TEI Consortium. https://tei-c.org/Vault/P5/4.9.0/doc/ tei-p5-doc/en/html/CO.html#COED. Can we make an AI respect TEI XML? An Experiment with a Small-Scale Explainable AI Model Elisa Beshero-Bondar, Hadleigh Bills, Alexander Fisher Penn State Erie, USA Keywords AI, LLM, TEI XML, XPath, RAG Abstract My students and I are developing a project to see whether we can refine and customize a “DigitAI” (named for our “Digit” program at Penn State Behrend). We want our DigitAI to serve as a guide to TEI as well as the XML stack technologies used in our program and projects. We want our AI to be more reliable than previous experiences with LLMs in coding assistance, because (if we are successful) it will deploy XPath and XML-processing technologies. The challenge of our project is to introduce the meaningful hierarchy of XML and TEI to an AI language model trained on sequenced tokens and word vector embeddings. If successful, the DigitAI will be minimal in size and adaptable by others, but most importantly, we hope to learn from the experiment. We are designing an “explainable AI” system that gives humanities students and scholars agency in the training, evaluation, and application of the AI system, heeding the call for 38 Can we make an AI respect TEI XML? explainable modeling work in the January 2023 special issue of IJDH (Ries). In attempting this, we are learning from two previous efforts shared at the TEI 2024 conference. Martina Scholger and Mohamed Khemakhem each presented papers about creating an AI assistant for encoding documents in TEI, and each made use of Retrieval Augmented Generation or RAG to construct a local knowledge base for accessible to a large language model. Scholger spoke of delivering specialized resources from a scholarly edition project to enhance the knowledge base as well as prompt engineering strategies to improve results. Khemakhem supplied the TEI Guidelines as a document to augment a language model’s training, but found significant problems in the model’s responses due to the manner in which the Guidelines are delivered to the model via “chunking” their text—sometimes in the middle of a section or example. To address the problem that Khemakhem discussed at TEI 2024, we are investigating the parsing of XML as a resource for RAG. Thanks to Jang and Lee’s successful efforts in an architectural studies context, where generative AI models were able to work with XML input of BIM (building information modeling) data, we are optimistic that we can introduce the XML of the TEI Guidelines to a generative AI model, or at least to control the chunking of the input data by following element structures. We have chosen LLaMa, introduced by Meta in 2023, as a customizable small language model because its models are cross-platform compatible (Macs, PCs, Linux), provided that machines have sufficient GPU and 16 GB of RAM. We are experimenting with putting lxml etree and possibly SaxonC libraries into the Python script we are using to customize our model. We may be deploying XSLT in our pipeline to output digestible and coherent units of information to be introduced as a RAG resource. By the time of the TEI conference, I hope to report on our progress, share how far we have succeeded and/or failed in our efforts, and what we have learned so far. References Hassan El-Hajj, Oliver Eberle, et. al. “Explainability and transparency in the realm of digital humanities: toward a historian XAI.” International Journal of Digital Humanities (IJDH) (2023) 5:299-331. https://link.springer.com/article/10.1007/s42803023-00070-1 Accessed 2024-12-08. Khemakhem, Mohamed et. al. “Enhancing Technical Knowledge Acquisition with RAG Systems: The TEI Use Case.” TEI 2024 Conference presentation at Buenos Aires. 2024. https://zenodo.org/records/13988319 Accessed 2024-12-08. Jang, Suhyung and Ghang Lee. Interactive Design by Integrating a Large Pre-Trained Language Model and Building INformation Modeling” Computer Science > Artificial Intelligence. arXiv: https://arxiv.org/abs/2306.14165 Llama Hub resources for building RAG applications. https://llamahub.ai/ . Accessed 2024-12-08. Newtfire.org training resources. https://newtfire.org/ . Accessed 2024-12-08. Ries, Thorsten, Karina van Dalen-Oskam, and Fabian Offert. “Reproducibility and explainability in digital humanities.” International Journal of Digital Humanities (IJDH). (2024) 6: 1-7. https://doi.org/10.1007/s42803-023-00083-w Accessed 2024-12-08. Scholger, Martina et. al. “Empowering Text Encoding with Large Language Models: Benefits and Challenges.” TEI 2024 Conference presentation at Buenos Aires. 2024. https://zenodo.org/records/13969082 Accessed 2024-12-08. TEI Consortium, eds. Guidelines for Electronic Text Encoding and Interchange. TEI P5 Version 4.8.1. Last modified 1 November 2024. https://tei-c.org/release/doc/tei-p5-doc/en/html/index. html. 40 Converting data from Czech digital libraries to TEI Converting data from Czech digital libraries to TEI 41 LONG PAPER Converting data from Czech digital libraries to TEI Boris Lehečka Moravian Library in Brno, Czech Republic Keywords digital library, natural language processing, text enrichment, ALTO Abstract In the years 2019-2022, within the project „DL4DH – Development of tools for more effective use and mining of digital library data to strengthen digital humanities research“, a set of applications has been developed that uses (meta)data stored in the Kramerius system, designed for Czech digital libraries, and converts them (i.e. ALTO) into TEI format, while the text is enriched using external tools UDPipe and NameTag. UDPipe identifies sentences in the text and tokens within them, which it lemmatizes and annotates with morphological annotation, for 80 languages. NameTag tokenizes the text and recognizes named entities in it: persons, geographical places, institutions, time and numerical data, etc., for 20 languages. The resulting TEI document is created by taking bibliography from MODS file and by merging three documents from partial transformations (from ALTO, NameTag and UDPipe). The <facsimile> element identifies the recognized areas (<zone>) in the original image. The text is divided into paragraphs (<p>) and sentences (<s>), individual words (<w>) use attributes (@lemma, @msd) for lemmas and morphological categories. The header of the final document contains the categories of the recognized entities (in the <textClass> element) as well as information about the tools used during the enrichment: name of the application, the language model used, and the licensing arrangements for the output (<appInfo> and <availability>). The first problem in transforming and merging data arises from the fact that in the ALTO format, the input data is divided into blocks and lines but not into paragraphs, which makes it difficult to split the text into separate sentences for further analysis. ALTO may contain recognition errors (at the character level), which complicates lemmatization and morphological analysis. The first problem can be addressed by analyzing the input blocks (e.g., values for indentation of the first line or the whole block, first uppercase letter, hyphens at the end of lines, etc.) and grouping related lines into a paragraph. Another problem is that the external applications UDPipe and NameTag work not with XML on the input, but only with plain text (although it can have a structured form, e.g. vertical, CoNLL-U, etc.). The output is available as XML (for NameTag) or as structured plain text (NameTag, UDPipe). In order to avoid problems with merging the output data because of different numbers of tokens, the UDPipe service is first used to split the input text into tokens and to apply morphological analysis and lemmatization. The output in CoNLL-U format is later used as input for entity recognition. When sending the data, it is also advisable to provide the external tools with information about the language of the analyzed text. This can be done using metadata about the publication or recognized language recorded in the ALTO format. Complications arise when there are isolated foreign expressions within a paragraph. Within the Libri augmentati application (programmed in XProc 3.1 with individual researchers in mind), a set of XSLT and XQuery 42 Converting data from Czech digital libraries to TEI transformations provides the conversion of input data (ALTO, NameTag XML, or CoNLL-U) into TEI format. References Boris Lehečka, (2025). Libri augmentati [online]. Version 1.0.0. [computer software]. Brno: Moravian Library in Brno. [Viewed 11 July 2025]. Available from: https://github.com/ moravianlibrary/libri-augmentati Lehečka, B., Novák, D., Kersch, F., Hladík, R., Bíšková, J., Sekyrová, K., Válek, F., Vozár, Z., Bodnár, N., Sekan, P., Bežová, M., Žabička, P., Lhoťák, M. and Straňák, P., (2022). Metodika přípravy dat z digitálních knihoven pro využití v digitálních humanitních vědách [online]. Knihovna AV ČR, v. v. i. [Viewed 11 July 2025]. Available from: https://hdl.handle.net/11104/0335692 Library of Congress, (2023). ALTO: Analyzed Layout and Text Object [online]. ALTO: Technical Metadata for Layout and Text Objects (Standards, Library of Congress). [Viewed 11 July 2025]. Available from: https://www.loc.gov/standards/alto/ Straka, M., (2018). UDPipe 2.0 Prototype at CoNLL 2018 UD Shared Task. In: Proceedings of CoNLL 2018: The SIGNLL Conference on Computational Natural Language Learning [online]. Stroudsburg, PA, USA: Association for Computational Linguistics. pp. 197–207. [Viewed 11 July 2025]. Available from: https://aclanthology.org/K18-2020/ Straková, J., Straka, M. and Hajič, J., (2019). Neural Architectures for Nested NER through Linearization. In: Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics [online]. Stroudsburg, PA, USA: Association for Computational Linguistics. pp. 5326–5331. [Viewed 11 July 2025]. Available from: https://aclanthology.org/P19-1527/ TEI Consortium, (2025). TEI P5: Guidelines for Electronic Text Encoding and Interchange [online]. The Text Encoding Initiative Consortium. [Viewed 11 July 2025]. Available from: https:// www.tei-c.org/Vault/P5/4.9.0/doc/tei-p5-doc/en/html/ Walsh, N., Berndzen, A., Imsieke, G. and Siegel, E., (2025). XProc 3.1 [online]. XProc - Specifications. [Viewed 11 July 2025]. Available from: https://xproc.org/specifications.html 44 Differentiating ODDs Differentiating ODDs Syd Bauman (1); Helena Bermúdez Sabel (2); Martin Holmes (3) (1) Northeastern University, USA; (2) JinnTec, Germany; (3) University of Victoria, Canada Keywords TEI, ODD, ATOP Abstract As we understand it, the original plan for the TEI/ODD system was for two types of user-facing ODD files: 7 base ODDs which contain specifications for classes, elements, attributes, datatypes, and macros, likely arranged within modules; and customization ODDs which assert which specifications from the associated base ODD should be used (possibly with additional specifications) to create a particular markup language. These two types were to be differentiated by the presence of a <schemaSpec> element in the latter, and not in the former. But three realities impinge on that simple plan: 1. Many users of the TEI/ODD system (including two of the authors) have created stand-alone markup languages (i.e., languages written in ODD but not based on TEI) using a single ODD file that has a <schemaSpec> element. Such an ODD file is not a customization ODD, but nonetheless has a <schemaSpec> 7 The ATOP project has developed a terminology system to help clarify several different types of ODD file, as well as the processes that take place during ODD processing; this can be found on the project Wiki: https://github.com/TEIC/atop/wiki/Terminology. During the presentation, we will briefly explain this terminology. SHORT PAPER 2. Using the current Stylesheets the “compiled” ODD (the initial output of combining a customization ODD with a base ODD) contains a <schemaSpec> element. This compiled ODD file may be used to generate a customized schema (in Schematron and either RELAX NG, XSD, or DTD), but also may be used as the base for another customization, in which case it is not a customization ODD, but nonetheless has a <schemaSpec>. 3. ODD-chaining (the process of creating a sequence of successive customizations, where each stage is turned into the base ODD for the next stage) has become prevalent and useful to the community. This means that any individual customization may itself become a base ODD, while retaining its <schemaSpec>. Thus we cannot use the mere presence of a <schemaSpec> to automatically differentiate between these two types of ODD files, which we need to be able to do to effect automatic ODD chaining. Our paper will discuss these issues, demonstrate a moderately complicated XSLT function that we think would do the job,8 and suggest a simple modification to the TEI scheme that would solve the problem: Making the @source attribute of <schemaSpec> a required attribute, and providing a special-purpose value (e.g. “tei:NONE”) for use when the ODD is a base ODD. 8 For the current version, see the atop:is-base-odd() function in XSLT/ modules/functions_module.xslt in the TEIC/atop git repository. 46 Dispatches from the TEI’s GIS Working Group Dispatches from the TEI’s GIS Working Group Joey Takeda (1); Martin Holmes (2) (1) Simon Fraser University, Canada; (2) University of Victoria, Canada Keywords GIS; geohumanities; interchange Abstract Created in November 2024, the GIS Working Group (GIS-WG) was charged by the TEI Technical Council with proposing modifications to the TEI schema to support more detailed, granular, and standards-based methods of encoding geospatial data in TEI, along with corresponding updates to the Guidelines prose. The full charge is available at https://tei-c.org/activities/ workgroups/gis-charge/. This presentation will share the current state of the GIS-WG’s work, outlining the group’s objectives, current recommendations, and the rationale behind them. Early in 2025, we devised a survey on current practices and preferences, which was completed by a number of community members particularly interested in this topic. We have also received feedback and discussion from the community during the March TEI Community Call. Drawing on community feedback and informed by existing practices, the GIS-WG has now developed a proposal for schema modifications allowing support for GeoJSON, WKT, GML, KML, and other standards within TEI encoding. We have sought SHORT PAPER to ensure that these proposed changes are sufficiently flexible to support widely-used standards, appropriately constrained to facilitate robust interchange, and backwards compatible with existing practices. We propose expanding <geo> to allow content in any of the supported geographic standards and creating a new element, <geoDef>, to allow for centralized definitions of geographic schemes. These two elements will both be members of a new class of attributes, which will specify the encoding scheme used, its version, and, to promote interchange, the level at which the geographic scheme is being implemented. 48 Editing Multilingual Grand Vizierial Correspondence in TEI Editing Multilingual Grand Vizierial Correspondence in TEI The GraViz project Stephan Kurz; Yasir Yılmaz; Dimitra Grigoriou; Nilab Saeedi; Michael Vogelsberger Austrian Academy of Sciences, Austria Keywords Early modern history, TEI edition, Multilingual, Non-Latin script, Ottoman Empire, Habsburg Monarchy Abstract This poster charts new territories: Ottomanists and historians have studied source material available in Turkish and European archives for a long time. While research has produced numerous historiographical outcomes that draw on those sources, the number of scholarly editions of sources remained low as a consequence of the dispersal of sources and the specific challenges of multilingual textual transmission. The limitations to access such resources include that most existing collections provide little to no open source data. The project ›The Ottoman Grand Vizierate (1560s to 1760s)‹ – GraViz in short – centers around the diplomatic correspondence of six selected grand viziers of the Ottoman Empire, preserved POSTER in the archives in Istanbul and Vienna. The project’s editorial mission is to apply state of the art TEI encoding to these letters, which survive in multiple languages: The GraViz team transcribes texts in Ottoman Turkish, Latin, and German language, and translates all Ottoman Turkish texts and a select group of German and Latin letters into English. Documents written in other languages, e.g. Italian and French, may follow in later stages, as well as detailed prosopographical information on the individual grand viziers. The editorial team references named entities (persons, places) and dates (challenges in premodern calendar systems ensue), and provides keywords and an English-language abstract. CorrespDesc and other editorial metadata are included in the editions, along with facsimile images that show the challenging layout and reading order. The project team published the first batch of edited documents, containing the correspondence of Sokollu Mehmed Pasha (in office: 1565–1579), at https://qhod.net/context:graviz?locale=en. The poster contribution focuses on how the GraViz project creates TEI XML data (schema, auxiliary standOff data, teiHeader, alignment of original and translation/s), which serve as a solid digital foundation for future research on textual artifacts of diplomacy between the Ottoman Empire and European courts. Encoding language diversity in TEI 63 DLA Marbach, ed. 2015. “Hilfe!” Wiki des Projekts ‚Netzliteratur authentisch archivieren und verfügbar machen.‘ Last changed Mai 3, 2016, accessed April 29, 2025. https://wwik.dla-marbach.de/line/index.php?title=Hilfe!&oldid=4070. Landow, George P. 1992. Hypertext. The Convergence of Contemporary Critical Theory and Technology. Baltimore: Johns Hopkins University Press. Short, Emily. 2024. “Narrative and Interactivity.” In: The Cambridge Companion to Literature in the Digital Age, edited by Adam Hammond, 177–193. Cambridge: Cambridge University Press. https://doi.org/10.1017/9781009349567.012. TEI Consortium. 2025. “Computer-mediated Communication.” In: TEI: Guidelines for Electronic Text Encoding and Interchange. P5 Version 4.9.0, 24th January. https://tei-c.org/release/doc/ tei-p5-doc/en/html/CMC.html. Encoding language diversity in TEI A description of regional and nonstandard languages in multilingual diachronic corpora Juliette Janès; Rachel Bawden; Thibault Clérice; Rasul Dent; Lucence Ing; Oriane Nédey; Benoît Sagot INRIA, Paris, France Keywords TEI ODD, fine-grained language description, non-standard languages, multilingual corpora Abstract Currently, French minority and regional languages lack digital resources, especially for cultural preservation and equitable representation. With the emergence of new tools for speakers, it is crucial to build stable corpora for under-resourced languages. In France, initiatives like DIVITAL9 and the earlier RESTAURE project10 have developed resources for specific languages, but much remains to be done. 9 https://divital.gitpages.huma-num.fr/fr/ 10 https://restaure.unistra.fr/ SHORT PAPER 64 Encoding language diversity in TEI Encoding language diversity in TEI 65 To address this gap, the COLaF11 project aims to develop corpora and Natural Language Processing (NLP) tools for the languages of France, spanning different languages and dialects, time periods, geographic regions and social contexts. Even within one language, variation happens depending on temporal, geographical and sociolinguistic factors (Labov, 1989), especially for non-standard languages. While NLP increasingly addresses linguistic diversity (Broeder, 2006) and TEI elements exist for this purpose, to our knowledge, no previous project has tackled the issue of a standardized structure for linguistic variation and its associated parameters in TEI corpora. We propose a TEI schema to enable the fine-grained description of language variation, with a particular focus on non-standard languages. 12,13 We use a combination of <language>, <particDesc> and <settingDesc> elements to document the manifestation of a language as it is spoken in a specific place, at a specific time and within a particular context. As BCP47 does not include all the ‘regional languages’ we would like to work on, we use Glottolog identifiers (Nordhoff, 2011), for a wider range of languages. We will later propose various language classifications to browse the corpus to address the multiple representations of languages and dialects. We specify both elements to include detailed metadata on the language, individuals associated with the document, and collection location, using controlled vocabularies where possible 14 (Figures 1, 2 and 3). 11 COLaf (‘Corpus et Outils pour les Langues de France) ‘Corpus and Tools for the Languages of France’), is an ongoing project funded by Inria to contribute to the development of free corpora and tools for French and other languages of France. For further information see: https://colaf.huma-num.fr/en/ 12 The ODD schema and document is available here: https://github. com/DEFI-COLaF/metadata 13 The ODD has been already used for various multi-speaker corpora such as the Molye (https://github.com/DEFI-COLaF/Molye), Concours Picard (https://github.com/DEFI-COLaF/Datasets_text/tree/main/ Picard_Concours) and Forum Occitania corpora. 14 For example, Glottolog identifiers for the language id, Geonames identifiers (https://www.geonames.org/) for the place id, etc. The metadata is linked to the text via attributes @who, @corresp, @hand, and @xml:lang (Figure 4). Fig. 1. Example of a content for a <language> element Fig. 2: Example of a content for a <person> element 66 Encoding language diversity in TEI Fig. 3. Example of a content for a <place> element Fig. 4. Example of a linked content References Broeder, Daan, et Peter Wittenburg. « The IMDI metadata framework, its current application and future direction ». Int. J. Metadata Semant. Ontologies 1, no 2 (1 octobre 2006): 11932. https://doi.org/10.1504/IJMSO.2006.011008. Labov, William. Sociolinguistique. Traduit par Alain Kihm. 1 vol. Le sens commun. Paris: Minuit, 1989. Nordhoff, Sebastian, et Harald Hammarström. « Glottolog/ Langdoc: Defining Dialects, Languages and Language Families as Collections of Resources ». In Proceedings of the First International Workshop on Linked Science 2011, 783:7. Bonn, Germany: CEUR, WS, 2011. https://ceur-ws.org/Vol-783/paper7.pdf. TEI Consortium, eds. “2.4.2. Language Usage” TEI P5: Guidelines for Electronic Text Encoding and Interchange. [2025-01-24]. https://tei-c.org/release/doc/tei-p5-doc/fr/html/ HD.html#HD41 (2025-04-07) TEI Consortium, eds. “16.2 Contextual Description” TEI P5: Guidelines for Electronic Text Encoding and Interchange. [2025-01-24]. https://tei-c.org/release/doc/tei-p5-doc/en/ html/CC.html#CCAH (2025-04-07) 68 Encoding Ligature Glyphs in Arabic Manuscripts Encoding Ligature Glyphs in Arabic Manuscripts Using TEI-XML Serhat Acar UBFC, Germany Keywords Arabic script encoding, Ligatures in Arabic script, TEI customization Arabic Abstract Encoding right-to-left scripts in TEI-XML, specifically Arabic, presents significant technical challenges, particularly in represent - ing ligature glyphs and complex character components. Arabic script is characterized by its use of numerous ligatures—combinations of characters that are written as a single, unified glyph. Properly encoding these ligatures is essential for maintaining the integrity and readability of the manuscripts. A primary issue is that standard TEI-XML tags and attributes may not sufficiently capture the nuances of these ligatures. The TEI framework provides the <g> element to specify individual glyphs or character components, ensuring accurate representation and searchability. For instance, an Arabic ligature such as “لا” (lam-alif) can be encoded as: <p><g ref=”#ligature_arabic_lam_alef”> لا </g> يمكن استخدام النص العربي المتصل . </p> This approach requires a detailed markup scheme, and a comprehensive understanding of the specific ligatures used in the SHORT PAPER manuscript. This can complicate the encoding process, especially for editors who may not be familiar with all the nuances of Arabic script. Accurately representing these ligatures involves ensuring that the TEI-XML documents maintain the correct rendering of text flow and character combination, as managed by the Unicode Bidirectional Algorithm. This is crucial for preserving the original format and readability of the manuscripts. For example, managing punctuation and embedded Latin text within Arabic script requires careful handling to ensure proper display and interpretation: <pb n=”3” facs=”arabic-script-page3”> <bidiOverride dir=”rtl”> <p> مثال على النص العربي مع تضمين <hi rend=”italic”>Latin</hi> نص .</p> </bidiOverride> </pb> This proposal aims, to focus on the challenges and solutions related to encoding ligature glyphs and complex character components in Arabic manuscripts using TEI-XML. By leveraging the TEI framework’s capabilities, it is possible to create a robust encoding scheme that faithfully represents Arabic ligatures, enhancing the accessibility of these texts for scholarly research. References Unicode Consortium. (2021). Unicode Bidirectional Algorithm. Retrieved from https://www.unicode.org/reports/tr9/tr946.html TEI Consortium. (n.d.). TEI Guidelines. Retrieved from http://www. tei-c.org/Guidelines/ 70 Encoding Multilingual and Multiscript Sources in TEI… Encoding Multilingual and Multiscript Sources in TEI… 71 Encoding Multilingual and Multiscript Sources in TEI A historical Spanish-Chinese dictionary Martina Scholger; Elisabeth Steiner; Sabrina Strutz; Melanie Frauendorfer University of Graz, Austria Keywords multilingual, multiscript, digital scholarly edition, historical lexicography, tonal languages Abstract The importance of documenting yet under-resourced language varieties and strengthening the collaboration between digital philology and language documentation has gained increasing recognition (Bowers, 2020; Czaykowska-Higgins et al., 2014; Ngué Um, 2017; Thompson, 2023; VedaWeb, 2024). In this context, we present our approach to encoding a multilingual, multiscript historical dictionary using the TEI standard. We focus on addressing key challenges, including orthographic variation, the representation of tonal features, and the need to balance historical fidelity with the requirements of contemporary research. The Early Manila Hokkien (EMHo, Döhla et al., 2024, Döhla et al., 2022) project digitizes and analyzes the 17th-century “Bocabulario de lengua sangleya por las letraz de el A.B.C.” (henceforth Bocabulario, British Library Add ms. 25.317), a Chinese-Spanish dictionary documenting the Hokkien variant LONG PAPER spoken by Chinese immigrants in early Manila. It serves as a valuable witness to historical language contact phenomena and provides unique insights into the massive linguistic and cultural exchange involving multiple languages and ethnic groups during this period in the Philippines, in particular the Spanish-Chinese relations (Döhla, 2025). A fundamental tension in our multi-layered encoding approach lies between diplomatic accuracy to the historical source, and the creation of structured research data that meets modern scholarly requirements and computational tractability. This ‘representational gap’ manifests in several ways. For example, the Bocabulario contains inconsistent orthography and spacing, requiring careful decisions about the depth of normalization. Fig. 1: Add ms. 25.317, fol. 2v. Facsimile reused with kind permission © British Library Board, all rights reserved. The data model is based on the Lex-0 TEI dictionary encoding recommendations (Tasovac et al., 2018). A simple dictionary entry (Fig. 1) includes a lemma (<form>), a word sense group (<sense>), a definition (<def>), and examples (<cit>) in Hokkien, along with Spanish translations. We use <entryFree> to preserve the original spelling, punctuation, and structure for diplomatic transcription, 72 Encoding Multilingual and Multiscript Sources in TEI… Encoding Multilingual and Multiscript Sources in TEI… 73 while editorial interventions and notes enable the creation of a normalized and critical edition. The <entry> element expands the structure beyond the historical source by incorporating expert-supplied modern Taiwanese romanization, Chinese characters, and English translations. The markup addresses the multilingual and multiscript challenges by applying the @xml:lang attribute to identify language varieties and their script representations, following the IETF BCP 47 standard (Philipps and Davis, 2009). The constant switching requires meticulous attention to markup consistency. In addition, we implement a linguistic analysis layer that integrates interlinear glossing following the Leipzig Glossing Rules (EVA MPG, 2015). The project aims to contribute methodologically to the TEI Guidelines by developing a recommendation for encoding tonal features which occur in approximately 70% of the world’s languages (Yip, 2002; Maddieson, 2013). We particularly emphasize interoperability to enable cross-domain reuse of this unique historical source by both linguists and historians. Thus, we aim to bridge the gap between multiple disciplines and address the challenges arising from that. Furthermore, this work contributes to broader efforts to document multilingual language contact scenarios. References Bowers, J. (2020) ‘Language documentation and standards in Digital Humanities: TEI and the documentation of MixtepecMixtec’. Computation and Language [cs.CL]. École Pratique des Hauts Études. https://tel.archives-ouvertes.fr/tel-03131936. Carroll, S. R., Garba, I., Figueroa-Rodríguez, O. L., Holbrook, J., Lovett, R., Materechera, S., Parsons, M., Raseroka, K., Rodriguez-Lonebear, D., Rowe, R., Sara, R., Walker, J. D., Anderson, J. and Hudson, M. (2020). ‘The CARE Principles Code Example 1: Diplomatic transcription, critical edition, and translation of one dictionary entry. Simplified and shortened. 74 Encoding Multilingual and Multiscript Sources in TEI… for Indigenous Data Governance’. Data Science Journal, 19: XX, pp. 1–12. DOI: https://doi.org/10.5334/dsj-2020-043. Czaykowska-Higgins, E., Holmes, M. D. and Kell, S. M. (2014) ‘Using TEI for an Endangered Language Lexical Resource: The Nxaʔamxcín Database-Dictionary Project’. Language Documentation & Conservation 8: 1 –3 7. Döhla, H. (2025) ‘The Bocabulario de la Lengua Sangleya por las Letras de el A.B.C. (Manila, ca. 1617): structure, contents, perspectives’. Historiographia Linguistica. https://doi.org/10.1075/ hl.00170.doh. Döhla, H., Klöter, H. and Scholger, S. (eds.). (2024) ‘Early Manila Hokkien. Digital Edition of the Bocabulario de lengua sangleya’. GAMS. https://gams.uni-graz.at/emho. Döhla, H., Klöter, H., Scholger, M. and Steiner, E. (2022) ‘Annotating a historical manuscript as a linguistic resource’. TEI2022 Conference Book (1.1). TEI2022 - “Text as Data” (TEI2022), edited by James Cummings. Newcastle, UK. Zenodo. https://doi. org/10.5281/zenodo.7120027, p. 129. EVA MPG (= Max Planck Institute for Evolutionary Anthropology, Department of Linguistics). (2015). ‘The Leipzig Glossing Rules: Conventions for interlinear morpheme-by-morpheme glosses’. https://www.eva.mpg.de/lingua/pdf/Glossing-Rules.pdf. Maddieson, I. (2013) ‘Tone’, in Dryer, M. S. and Haspelmath, M. (eds.) The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. http:// wals.info/chapter/13. Ngué Um, E. (2017) ‘Issues in digital text representation, online dissemination, sharing and reuse for African tone languages’. Proceedings of the 2nd Workshop on the Use of Computational Methods in the Study of Endangered Languages, Honolulu, Hawai‘i, 24–32. https://aclanthology.org/W17-0104.pdf. Phillips, A. and Davis, M. (eds.). (2009) ‘Tags for Identifying Languages’, BCP 47, RFC 5646, DOI 10.17487/RFC5646, https://www.rfc-editor.org/info/rfc5646. Tasovac, T., Romary, L., Banski, P., Bowers, J., de Does, J., Depuydt, K., Erjavec, T., Geyken, A., Herold, A., Hildenbrandt, V., Khemakhem, M., Lehečka, B., Petrović, S., Salgado, A. and Witt, A. (2018) ‘TEI Lex-0: A baseline encoding for lexicographic data’. Version 0.9.3. DARIAH Working Group on Lexical Resources. https://dariah-eric.github.io/lexicalresources/ pages/TEILex0/TEILex0.html. Thompson, W. (2023) Epifanii Slavinetskii’s Greek–Slavonic–Latin Lexicon Between East and West. PhD Thesis, Heidelberg. TEI Consortium (eds.). (2025) ‘10 Dictionaries’. TEI P5: Guidelines for Electronic Text Encoding and Interchange. 4.8.1. TEI Consortium. https://tei-c.org/Vault/P5/4.8.1/doc/tei-p5-doc/ en/html/DI.html. VedaWeb. (2024) ‘Online Research Platform for Old Indic Texts’. https://vedaweb.uni-koeln.de. Yip, M. (2002). Tone. Cambridge Textbooks in Linguistics. Cambridge: Cambridge University Press. https://doi. org/10.1017/CBO9781139164559. 76 Enriching textual data from digital libraries Enriching textual data from digital libraries 77 Enriching textual data from digital libraries Boris Lehečka Moravian Library in Brno, Czech Republic Keywords digital library, natural language processing, text enrichment Abstract The demonstration will introduce open-source tools that can mine and enrich data from Czech digital libraries that use the Kramerius system. In 2019-2022, the DL4DH project (Development of tools for more effective use and mining of digital library data to strengthen digital humanities research) has developed a set of applications, DL4DH Kramerius+, Feeder and TEI Converter, that allow researchers to retrieve data and metadata form digital library not only in existing original formats (FOXML, MODS, ALTO, plain text, JPG), but also in TEI format, which can contain morphological data and recognized entities in addition to the text itself. Data are enriched by external tools, UDPipe and NameTag and transformed into the TEI format. Based on a contract between the library and the research institution (universities, institutes of the Academy of Sciences, etc.), researchers can request the enrichment of specific titles and download them for research, e.g. for training artificial intelligence. The demonstration will also include the Libri augmentati software application. Like the server applications from the DL4DH project, it serves the same purpose, but it is programmed in DEMONSTRATION XProc 3.0 and can run on desktop computers running Windows, macOS or Linux. It is intended for individual researchers and for processing smaller volumes of publications, especially freely available (in Czech digital libraries monographs published before 1950, periodicals before 1910). The application comprises a set of modules: for enrichment text from ALTO format or plain text using external tools, UDPipe and NameTag; all data are at the end converted to TEI format. The modules can be developed to be more generic so that they can be used not only with Kramerius digital libraries but also with digital libraries that use different APIs and metadata, such as IIIF. References Knihovna Akademie věd ČR, Národní knihovna ČR, Moravská zemská knihovna v Brně and InQool, (2022a). DL4DH Kramerius+ [online]. Version 0.10.0. [computer software]. [Viewed 11 July 2025]. Available from: https://github.com/ LIBCAS/DL4DH-Kramerius-plus Knihovna Akademie věd ČR, Národní knihovna ČR, Moravská zemská knihovna v Brně and InQool, (2022b). DL4DH TEI Converter [online]. Version 0.6. [computer software]. [Viewed 11 July 2025]. Available from: https://github.com/LIBCAS/ DL4DH-TEI-Converter Knihovna Akademie věd ČR, Národní knihovna ČR, Moravská zemská knihovna v Brně and InQool, (2023). DL4DH Feeder [online]. Version 1.3.0. [computer software]. [Viewed 11 July 2025]. Available from: https://github.com/LIBCAS/ DL4DH-Feeder Boris Lehečka, (2025). Libri augmentati [online]. Version 1.0.0. [computer software]. Brno: Moravian Library in Brno. [Viewed 11 July 2025]. Available from: https://github.com/ moravianlibrary/libri-augmentati Exploring correspSearch v3 as a Service for Minimal Editions… 79 Lehečka, B., Novák, D., Kersch, F., Hladík, R., Bíšková, J., Sekyrová, K., Válek, F., Vozár, Z., Bodnár, N., Sekan, P., Bežová, M., Žabička, P., Lhoťák, M. and Straňák, P., (2022). Metodika přípravy dat z digitálních knihoven pro využití v digitálních humanitních vědách [online]. Knihovna AV ČR, v. v. i. [Viewed 11 July 2025]. Available from: https://hdl.handle.net/11104/0335692 Straka, M. and Straková, J., (2025). NameTag [online]. Version 3.1.0. [computer software]. Praha: LINDAT/CLARIAH-CZ, digitální knihovna při Ústavu formální a aplikované lingvistiky, Matematicko-fyzikální fakulta Univerzity Karlovy. [Viewed 11 July 2025]. Available from: http://hdl.handle. net/11858/00-097C-0000-0023-43CE-E Straka, M. and Straková, J., (2023). UDPipe [online]. Version 2.1.0. [computer software]. Praha: LINDAT/CLARIAH-CZ, digitální knihovna při Ústavu formální a aplikované lingvistiky, Matematicko-fyzikální fakulta Univerzity Karlovy. [Viewed 11 July 2025]. Available from: http://hdl.handle.net/11234/1-1702 Exploring correspSearch v3 as a Service for Minimal Editions of Correspondence Stefan Dumont (1); Sascha Grabsch (1); Jonas Müller-Laackman (2); Ruth Sander (1); Steven Sobkowski (1) (1) Berlin-Brandenburg Academy of Sciences and Humanities, Germany; (2) State and University Hamburg Carl von Ossietzky Keywords correspondence, minimal edition, sustainability, web service, interoperability Abstract Since 2014, correspSearch aggregates the metadata of edited letters provided in the Correspondence Metadata Interchange Format (CMIF, developed by TEI Correspondence SIG) from different projects. With version 3 of correspSearch, a full-text search in the edited letters is now possible for the first time. To enable this, a URL to the corresponding TEI-XML file must be specified for each letter in the CMIF file, which correspSearch then retrieves. The TEI structure is also taken into account: letter text, editor’s comments and greeting/closing formulae can be distinguished. Additionally correspSearch v3 offers also different visualisations of search results or individual CMIF files: timeline of the correspondence(s), geographical distribution of sending and receiving places and a network visualisation. LONG PAPER 80 Exploring correspSearch v3 as a Service for Minimal Editions… Exploring correspSearch v3 as a Service for Minimal Editions… 81 Furthermore, it has recently become possible to search and filter for persons mentioned in the letters. The new functions also support the editing projects themselves. For example, the visualisations offer a graphical overview of the project’s data, which is often not available in the project itself. Some edition projects have therefore integrated these visualisations into their own digital offerings. One example of this is the Graz Nunciature Reports.15 The Buber Korrespondenzen digital16 goes a step further: The project, which runs until 2045, does not yet have a web platform, although letters edited in TEI-XML are already available. The project currently publishes these in a Gitlab repository. With the help of TEI Boilerplate, an HTML view is available on Gitlab pages. 17 This web presentation is referenced from the CMIF metadata. As a result, there is a human-readable view for each individual letter, but access in the sense of an overview with search and filter options is currently only offered via correspSearch.18 Both use cases lead to the question of whether digital editions of correspondence could be made more sustainable if correspSearch would be included as part of their conception. Common solutions for digital editions currently require a database and technologies based on it. However, these need to be maintained and updated even after the project has been completed. A more sustainable approach would be to publish the digital edition as a TEI-XML dataset and a static HTML website (generated via XSLT, TEI Boilerplate etc.). Both would require no further maintenance. The integration of correspSearch could 15 https://grazer-nuntiatur.acdh.oeaw.ac.at/visualizations.html 16 http://buber-korrespondenzen.digital/ 17 The presentation view with TEI Boilerplate is currently temporarily deactivated because it is being adapted more closely to the needs of the project. 18 https://correspsearch.net/de/suche. html?e=ae451a0a-2186-450c-ae5d-72024981937c&x=1&w=0 replace the need for a database and dynamic programming by offering a full-text search with extensive filtering options as well as visualisations. The prerequisites for this are already well laid out in correspSearch: The CMIF files as well as the TEI XML files for the fulltext search, can be stored permanently and statically on a web space or in a repository without the need for dynamic APIs. It also does not matter which technology was used to create the TEI XML data or the static HTML. The possibility of creating digital letter editions with manageable effort and without permanent maintenance obligations means that correspondence for which no larger DH resources can be provided could be published digitally. Often this concerns letters from non-canonised writers (such as women). With this in mind, our lecture will present the outlined possibilities and also discuss different types of integration, including their advantages and disadvantages in the context of ‘sustainability’ and ‘minimal computing / editions’. A specially created practical example will illustrate the interaction of correspSearch and static digital letter editions in the lecture. References Calarco, Gabriel A. 2023. ‘La écfrasis en el Libro de Alexandre, un proyecto de edición digital para el estudio de la poesía clerical castellana del siglo XIII con minimal computing’. Journal of the Text Encoding Initiative, no. Issue 16 (May). https://doi. org/10.4000/jtei.4494. Crompton, Constance. 2023. ‘“No Boutique or Fashionable Technologies”: Project Development, Mentorship, and Sustainability in an Innovation-First World’. Digital Humanities Quarterly 17 (1). https://digitalhumanities.org/dhq/ vol/17/1/000660/000660.html. 94 Foregrounding Text Analysis in Digital Scholarly Editing Foregrounding Text Analysis in Digital Scholarly Editing 95 to the text in the first place? Must we wait until after we have completed all the required editorial pieces? How can we leverage our encoding to pursue our questions and integrate our textual analysis into the process without impeding the needs of our publisher? And how important is it for our encoding work to appear in the final published edition? Can we not encode for our own needs at the same time that we are encoding for our (imagined) readers? In this talk I hope to propose that by building intentional phases for textual analysis into our editing process, we enrich the ultimate published edition, even if the published artifact doesn’t ‘show’ our textual analysis. Whether it be ‘seeing through’ our text in Voyant, or shifting from code-centric Oxygen to the text-forward LEAF-Writer environment, or revealing the patterns in our tagging in tools like the Dynamic Table of Contexts, if we can take the time to examine our work while we are editing it from different vantage points I suggest that we we can produce richer and more complex editorial work that what seems like an editorial luxury in fact makes us better editors. I will use my work on Shakespeare and Fletcher’s Henry VIII or All is True for the New Internet Shakespeare Editions22 to model different ways of intentionally integrating textual analysis into the encoding process, drawing on my own research questions about the play. My questions about royal identity, Henry VIII as a London City Play, and echoes in the play of Tudor propaganda for a playgoing audience in James I’s England have changed radically as a result of the sidebar text analysis. Ultimately, I’d like to encourage a discussion about how we as editors can give ourselves the opportunity to pause the publishing process and dive into encoding leveraged analysis at the same time that we honour the commitments we have made to publish our work. 22 With gratitude for patience (and apologies for tardiness) to the general editors of the New Internet Shakespeare Editions, which was still the Internet Shakespeare Editions when I first submitted my edition proposal. References NB: these authors and collaborators have greatly informed my thinking in this way. While they do not necessarily appear by name in the abstract, their work is in my head and I hope to draw much of it out more explicitly in my paper. Brown, Susan, Brent Nelson, Stan Ruecker, Stéfan Sinclair, Nadin Adelaar, Ruth Knechtel, and Jennifer Windsor. 2013. “Text Encoding, the Index, and the Dynamic Table of Contexts.” at the Annual Meeting of the Alliance of Digital Humanities. Lincoln, Nebraska. Cummings, James. 2019. “Opening the Book: Data Models and Distractions in Digital Scholarly Editing.” International Journal of Digital Humanities 1 (2): 179–93. https://doi.org/10.1007/ s42803-019-00016-6. Jakacki, Diane K. 2025. “Surfacing Encoding-Driven Analysis in Digital Editions with DToC.” At the Workshop GRENCRIHN « New Perspectives on Critical Editions » (Part 2). Montreal. https://www.crihn.org/nouvelles/2025/03/13/ workshop-new-perspectives-on-critical-editions-part-2/ —. 2023. Balancing Need, Speed, and the Future of Collaboration in Digital Scholarly Production.” Invited Talk. Digital Humanities Virtual Seminar hosted by THINC Lab (U Guelph), CRIHN (UdeM), The Humanities Data Lab (UOttawa), and the Centre for DH (Toronto Metropolitan University). —. 2018. “Internet Shakespeare Editions and the Infinite (Editorial) Others: Supporting Critical Tagsets for Linked Editions.” In Shakespeare’s Language in Digital Media: Old Words, New Tools, edited by Janelle Jenstad, Mark Kaethler, and Jennifer Roberts-Smith, 157–71. Routledge. —. n.d. “Henry VIII: Internet Shakespeare Editions.” Accessed April 27, 2025. https://internetshakespeare.uvic.ca/Library/ Texts/H8/. 96 Foregrounding Text Analysis in Digital Scholarly Editing Jakacki, Diane K. and Susan Brown. 2024. “‘I don’t even see the code:’ The Dynamic Table of Contexts – Optimizing our Encoded Texts for Web-based Reading and Analysis Environments.” TEI 2024 Annual Meeting. Jakacki, Diane K. and Janelle Jenstad. 2016. “Mapping Toponyms in Early Modern Plays with the Map of Early Modern London and Internet Shakespeare Editions Projects.” In Early Modern Studies and the Digital Turn, edited by Laura Estill, Diane Jakacki, Michael Ullyot, 237-258. ITER. Jenstad, Janelle. 2021. “Janelle Jenstad: Internet Shakespeare Editions - Speaking of Shakespeare.” SoS #16. https://www.buzzsprout.com/1732460/episodes/9123329-sos-16-janelle-jenstad-internet-shakespeare-editions. McGann, Jerome. 2014. A New Republic of Letters: Memory and Scholarship in the Age of Digital Reproduction. Cambridge, Massachusetts: Harvard University Press. —. 2001. Radiant Textuality: Literature After the World Wide Web. New York: St. Martin’s Press. Masten, Jeffrey. 2001. “More or Less: Editing the Collaborative.” Shakespeare Studies 29:109–31. Orley, Emily. 2022. “Editing as Creative Act: An Experiment in Speculative Thinking.” Textual Cultures 15 (1): 11–17. Pierazzo, Elena. 2016. “Modelling Digital Scholarly Editing: From Plato to Heraclitus.” In Digital Scholarly Editing, edited by Elena Pierazzo and Matthew James Driscoll, 1st ed., 4:41–58. Theories and Practices. Open Book Publishers. https://www. jstor.org/stable/j.ctt1fzhh6v.7. —. 2011. “A Rationale of Digital Documentary Editions.” Literary and Linguistic Computing 26 (4): 463–77. https://doi. org/10.1093/llc/fqr033. Rasmussen, Krista Stinne Greve. 2016. “Reading or Using a Digital Edition?: Reader Roles in Scholarly Editions.” In Digital Scholarly Editing, edited by Matthew James Driscoll and Elena Pierazzo, 1st ed., 4:119–34. Theories and Practices. Open Book Publishers. https://www.jstor.org/stable/j.ctt1fzhh6v.11. Turska, Magdalena, James Cummings, and Sebastian Rahtz. 2016. “Challenging the Myth of Presentation in Digital Editions.” Journal of the Text Encoding Initiative, no. Issue 9 (September). https://doi.org/10.4000/jtei.1453. Van Mierlo, Wim. 2022. “The Scholarly Edition as Digital Experience: Reading, Editing, Curating.” Textual Cultures 15 (1): 117–25. 98 From dialect features to structured data… From dialect features to structured data… 99 From dialect features to structured data Modelling spoken Arabic varieties in the WIBARAB project Veronika Engler (1); Ana Iriarte Díez (2); Karlheinz Mörth (1); Stephan Procházka (2); Laurent Romary (3); Daniel Schopper (1); Kinga Sramó (1); Katharina Wünsche (1) (1) Austrian Academy of Sciences, Austria; (2) University of Vienna, Austria; (3) National Institute for Research in Digital Science and Technology, France Keywords linguistics, oriental studies, language varieties, data modelling, visualisation Abstract The Vienna Corpus of Arabic Varieties (VICAV) is a language documentation platform that provides access to a growing collection of digital language resources. Integrating approaches from language technology and the wider field of text-oriented digital humanities, the project aims to address issues of representing heterogeneous data by providing a flexible, yet sustainable technical environment based on a modular data architecture. In addition to a bibliography of research literature, typologically similar to what can be found in the World Atlas of Language Structures (WALS) or the Database of Arabic Dialects (DAD), VICAV offers several types of data: the so far largest part of the LONG PAPER collection comprises linguistic profiles (i.e. standardised concise descriptions of linguistic varieties), structured lists of linguistic features, sample texts, corpora of unmonitored speech, and dictionaries. The VICAV infrastructure is meant to ensure consistent encoding across projects as well as sustainable creation and publishing workflows across projects and builds largely on TEI (P5) as its underlying data model (Procházka et al. 2015). The newest and, in terms of resources, duration, and scope, most extensive addition to the VICAV projects is the ERC Advanced Grant WIBARAB (What is Bedouin-type Arabic? 101020127-WIRARAB). It investigates the linguistic and socio-historical realities behind the millennia-old dichotomous distinction between Bedouin or sedentary dialects (Procházka 2024). The central component of WIBARAB is a linguistic feature database covering over 300 varieties of spoken Arabic. It incorporates data from hitherto little researched areas collected in campaigns in Saudi Arabia, Kuwait, Jordan, Lebanon, Sudan, Tunisia and Morocco, along with previously published material. The project has a strong focus on open access, data structures, standards and best practices, as well as data modelling. By contrast to comparable other projects collecting linguistic data, WIBARAB has adopted a strictly text-oriented approach that has been grounded in the application of the Guidelines of the Text Encoding Initiative (TEI). While TEI (P5) provides a welltried inventory of elements to represent morphological and lexical concepts, the issue of representing other grammatical phenomena has received less attention and modelling extra-linguistic and socio-linguistic phenomena constitutes the interesting part of this encoding challenge. Our paper will focus on the WIBARAB team’s customisation, trying to re-use as many concepts existing in the TEI guidelines as possible, document workflow steps such as the writing of a meaningful, re-usable ODD, and examine some encoding decisions by furnishing examples of phonological, morphological, syntactical, phraseological and lexical data. 100 From dialect features to structured data… Finally, we will touch on some methodological challenges in developing an efficient research-tool on the basis of this TEIencoded, intricately structured linguistic database. It has already been used for a wide range of varied research questions such as the in-detail description of particular linguistic varieties (Torzullo 2024) as well as socio-linguistic approaches dealing with questions of intergenerational variation, intra-speaker variation and identity (Iriarte Díez 2025a, 2025b). We will also provide a first glimpse of the evolving map-based front-end which will be integrated into the overall VICAV infrastructure. References Iriarte Díez, A. (2025a) ““Bedouins” in Beirut (and surroundings): Arab Khalde – Intergenerational linguistic variation in a traditionally seminomadic community 12 km south of Beirut”, in A. Iriarte Díez and St. Procházka (eds.) What is Bedouin-Type Arabic? New Data and Fresh Perspectives. Wiener Zeitschrift für die Kunde des Morgenlandes (WZKM) 115. Iriarte Díez, A. (2025b) “Intra-speaker variation and identity performance: Two texts in the Arabic of Arab Khalde (Lebanon)”, in A. Iriarte Díez and St. Procházka (eds.) What is Bedouin-Type Arabic? New Data and Fresh Perspectives. Wiener Zeitschrift für die Kunde des Morgenlandes (WZKM) 115. Procházka, S. (2024) “How solid is the linguistic basis for the Bedouin sedentary split used in the classification of Arabic dialects? ”, in C. B. Ramos, J. Guerrero and M. B. Fernández (eds.) AIDA Granada: A pomegranate of Arabic varieties. Zaragoza: Prensas de la Universidad de Zaragoza: 359-370. https:// phaidra.univie.ac.at/o:2108301. Procházka, S. and K. Moerth (2015) “The Vienna Corpus of Arabic Varieties: building a digital research environment for Arabic dialects”, in M. Al-Hamad, R. Ahmed and H. Aloui (eds.) Lisan Al-Arab: Studies in Contemporary Arabic Dialects, Proceedings of the 10th International Conference of AIDA, Qatar University 2013. Vienna: LIT Verlag: 209-218. Torzullo, A. (2024) “The Arabic Dialect of a Jordanian Camel-Breeder Tribe: a Comparative Analysis of Selected Phonological and Morphological Features of the Bani Ṣaxar Variety”, in C. B. Ramos, J. Guerrero and M. B. Fernández (eds.) AIDA Granada: A pomegranate of Arabic varieties. Zaragoza: Prensas de la Universidad de Zaragoza: 409-420. https:// phaidra.univie.ac.at/o:2108301. 102 From Practice to Framework. Model Digital Scholarly Editions… From Practice to Framework. Model Digital Scholarly Editions… 103 From Practice to Framework Model Digital Scholarly Editions in the Jagiellonian Digital Platform Magdalena Eulalia Komorowska; Iwona Grabska-Gradzińska; Joanna Hałaczkiewicz Jagiellonian University, Poland Keywords Digital Scholarly Editions, TEI XML, Digital Humanities Infrastructure, Jagiellonian Digital Platform Abstract Entering New Territories of Digital Scholarly Editing The Digital Scholarly Editing Laboratory (LabEdyt), an interdisciplinary team within the flagship project Digital Humanities Lab at the Jagiellonian University, was founded in 2022 to address the growing need for sustainable digital scholarly editions (DSEs) in Poland. While DSEs are already being developed with some regularity across many academic contexts, they remain a complex and evolving challenge – even in well-established environments. In the Polish context, their adoption is still limited to a few research centers and often regarded as too complicated to be worth considering. Persistent myths and misconceptions about the usefulness and durability of such editions, along with a reluctance to publish under open licenses, still act as a barrier to progress. LabEdyt’s mission is to create practical workflows . LONG PAPER and reusable models that lower the entry barrier to digital editing and support researchers in transitioning from print-based to digital methods. This paper presents the methodological framework and current outcomes of three model projects developed within the Jagiellonian Digital Platform. Mapping Workflows and Good Practices in Digital Editing The first project involves the revitalization of “Neolatina Sarmatica”, a digital collection of Latin texts by authors associated with the Polish-Lithuanian Commonwealth. Originally created between 2006 and 2011, the collection had become obsolete and lacked structured markup. By 2022, the existing website was no longer practical to use, as advances in web technologies had rendered it outdated. Such obsolescence of research websites – including digital scholarly editions – is a widespread problem. The pilot project on “Neolatina” therefore aimed not only to make this collection accessible to readers again, but also to develop guidelines for creating editions that remain sustainable over time and for reviving dormant digital projects. Using web scraping and TEI XML conversion, the team rebuilt the edition in TEI Publisher integrated with an external database. The new platform now supports side-by-side views of transcription, facsimile, and translation, along with interactive annotations. This project demonstrates the long-term value of adhering to encoding standards and underscores the importance of sustainable design for digital editions. The second project focuses on “Moralia”, a 17th-century manuscript by Wacław Potocki. The work of this Baroque poet comprises more than 2,100 pieces and was modeled on Erasmus of Rotterdam’s Adagia. Its considerable scale has long posed a challenge for editors, and it has never received a complete scholarly edition. The necessity of working with the manuscript also provided an opportunity to test a workflow incorporating a handwriting text recognition (HTR) tool. The team trained a custom handwriting recognition model in Transkribus on more 104 From Practice to Framework. Model Digital Scholarly Editions… than 100 annotated pages, achieving over 97% transcription accuracy. Current work involves developing a large language model capable of producing modernized versions of the text in line with the editorial principles. The model will be designed not only to ‘translate’ the diplomatic transcription into a modernized text, but also to automatically encode these changes in TEI. The edition will also integrate intertextual references, most notably a surviving copy of Erasmus’s Adagia annotated by Potocki himself. These materials will be re-framed in a digital commentary format through the use of structured markup and hypertext techniques. The third project aims to develop a method for constructing digital editions from materials originally prepared for print. As a case study, the team is working on a scholarly edition of 20th-century Szymon Laks’s correspondence with Krystyna and Czesław Bednarczyk, which had been typeset in Adobe InDesign and published in 2018. The project explores workflows for mapping style-based formatting onto XML structures. While certain structural elements can be retroconverted automatically, others – such as personal names or metadata – require manual tagging, custom scripting, or the use of AI-assisted tools like Named Entity Recognition. The challenges we continue to encounter reveal broader issues in digital retroconversion, particularly when editorial markup was not anticipated during the print-production stage. Building a Community of Practice All three projects are being released incrementally via the Jagiellonian Digital Platform (https://labedyt.dhlab.uj.edu.pl), which serves both as a stable public repository and as an internal development space for digital editors at the Jagiellonian University. Rather than inventing new proprietary tools, LabEdyt adapts and extends open-source solutions such as TEI Publisher × Jinks, contributing to their international development through applied use and structured feedback. In line with the “New Territories” theme, LabEdyt’s work represents an effort to establish digital scholarly editing as a viable and sustainable practice in Poland. By documenting challenges, sharing workflows, and publishing reusable models, the team supports a growing community of researchers who are ready to explore new directions at the intersection of textual scholarship and digital infrastructure. 106 Functors and format conversion Functors and format conversion 107 Functors and format conversion Riccardo Del Gratta; Angelo Mario Del Grosso Institute for Computation Linguistics A. ZampolliCNR-ILC, Italy Keywords Interoperability, Category Theory, Text Analysis, Computational Linguistics, LinkedOpen Data Abstract This contribution highlights the potential of functorial format conversion in applications such as digital philology and structured text analysis, where the heterogeneity of document formats often interferes with interoperability among text analysis tools. The authors introduce a formal framework based on Category Theory, using functors as an abstract means for managing format conversion. Functors maintain coherence during transformation, enabling tool pipelining even when format mismatches occur. In the proposed framework, functors enable the transformation of documents and their associated processing tools from one category (containing documents in format f) into another (in format g), preserving fundamental structural and operational properties. The idea is to represent format conversion not as a technical task but as a structured mapping that maintains the algebraic properties of the source and target domains. In previous works, the authors illustrated the practical implications of this approach through working examples that convert SHORT PAPER annotated XML into JSON documents, enriching them with externally linked data. This contribution underscores that beyond format conversion, the framework shows further research questions such as (i) document isomorphism; (ii) partial structure mapping, and (iii) natural transformations between functors. (i) and (ii) address the following scenario: First, an XML document D0_x is “functored” to a JSON document D0_j using functor F. Then, D0_j is “functored” into a possibly new XML document D1_x with functor G. In (i), we investigate suitable metrics to evaluate the similarity between D0_x and D1_x; in (ii), we study the implications of using functors that “forget” part of the original structure, i.e., the directive about the XML schema declaration. Finally, in (iii), we analyze the application of natural transformations to manage the mapping of D0_x into two JSON documents with different structures. (iii) is relevant to handle the “diffing” issues on documents that serialize the same content with different structures. References R. Del Gratta, F. Boschetti, L. Bambaci, and F. Sarnari, “Document analysis and Textual philology: A Formal Perspective,” International Journal of Information Science and Technology, vol. 5, no. 1, pp. 5–15, 2021. [Online]. Available: https://www.innove.org/ijist/index.php/ijist/article/view/192 –, “Approaching document analysis with a formal model,” in 6th International IEEE Colloquium on Information Science and Technology, Agadir, Morocco, 2020, pp. 208–214. R. Del Gratta, F. Boschetti, A. Del Grosso, S. Zenzaro, and L. Bambaci, “Philology as a dynamic system,” Umanistica Digitale, vol. 6, p. 1–20, Jan. 2022. [Online]. Available:https://umanisticadigitale.unibo.it/article/view/13684 Generating TEI Documents Through a Game of Dice 109 R. D. Gratta, S. Zenzaro, A. D. Grosso, and F. Boschetti, “Category theory, Document Analysis, and Philological Operations. A formal approach: limitations, and challenges,” International Journal of Information Science and Technology, vol. 9, no. 1, pp. 11–20, 2025. [Online]. Available: https://innove.org/ijist/index. php/ijist/article/view/288 S. Awodey, Category Theory, 2nd ed. New York, NY, USA: Oxford University Press, Inc., 2010. S. Mac Lane, Categories for the Working Mathematician, 2nd ed., ser. Graduate Texts in Mathematics. Springer, 1998. [Online]. Available: http://www.worldcat.org/isbn/0387984038 E. Riehl, Category Theory in Context, ser. Aurora: Dover Modern Math Originals. Dover Publications, 2017. [Online]. Available: https://books.google.it/books?id=6B9MDgAAQBAJ J. Gerrans and R. S. Sherratt, “Comparing xml and json characteristics as formats for data serialization within ultralow power embedded systems,” IEEE Embed. Syst. Lett., vol. 16, no. 4, p. 489–492, Dec. 2024. [Online]. Available: https://doi. org/10.1109/LES.2024.3450576 S. Zunke and V. D’Souza, “JSON vs XML: A comparative performance analysis of data exchange formats,” IJCSN International Journal of Computer Science and Network, vol. 3, no. 4, pp. 257–261, 2014. P. C. Lalith, S. Goel, M. Kakkar, and S. Sharma, “Simplifying code translation: Custom syntax language to c language transpiler,” in 2025 2nd International Conference on Computational Intelligence, Communication Technology and Networking (CICTN), 2025, pp. 1–6. A. Okutan, S. Merten, C. C. Michael, and B. Ryjikov, “Leveraging rag-llm to translate c++ to rust,” in 2024 International Conference on Assured Autonomy (ICAA), 2024, pp. 102–105. Generating TEI Documents Through a Game of Dice Viktor J. Illmer; Nele Heindorf; Roya Zendebudie; Mark Schwindt; Frank Fischer Freie Universität Berlin, Germany Keywords German Literature, Generative Literature, Drama, Gamification, 19th Century Abstract In 1829, German author Georg Nikolaus Bärmann published a 272-page book entitled “Neunhundert neun und neunzig und noch etliche Almanachs-Lustspiele durch den Würfel” (“Rolling the Dice for 999 and Many More Almanac Comedies”). The work consists of 1,200 randomly printed text pieces, mostly dialogs between the five main characters (the uncle, the niece, the maid, the lover, his servant). With the help of a dice, you can put together an entire German-language one-act comedy from these snippets. A table is provided for this purpose, in which a piece of text is assigned to each roll of a total of 200 rolls, from which a coherent play is then created. Mathematically, this results in 6 to the power of 200 possibilities and therefore more plays than atoms in the universe (Baker 2021). We first digitized Bärmann’s book using the “German Fraktur 18th Century – WrDiarium_M9” model on Transkribus and corrected it on Wikisource according to the four-eyes principle. We then produced a TEI version of the book featuring all 1,200 text snippets. On this basis, the book was turned into a web LONG PAPER 110 Generating TEI Documents Through a Game of Dice Generative AI for XML-TEI Encoding 111 app and the game can now be played online (Illmer et al. 2025). It is available at https://temporal-communities.github.io/999/. For the TEI2025 conference, we developed an extension to the app that allows you to download a randomly diced game in TEI format. This enables extensive analysis of the randomly generated plays. We present the TEI component of the game app as well as quantitative and stylometric experiments that we conducted with the generated plays. To facilitate the understanding of the play – written in a highly idiosyncratic German with historical orthography – we also provide an AI-assisted translation to English of the entire play. The translation is also encoded in TEI and can be accessed from within the app. To sum up, with this paper we will demonstrate how TEI can be used to encode a literary game and serve as a basis for an interactive web app, and we will discuss the implications for the TEI-encoding of AI-assisted translations. References Harry Baker: How many atoms are in the observable universe? In: Live Science, July 10, 2021. Available: https://www.livescience.com/how-many-atoms-in-universe.html Georg Nicolaus Bärmann: Neunhundert neun und neunzig und noch etliche Almanachs-Lustspiele durch den Würfel. Zwickau: Schumann 1829. (digitized version: https://de.wikisource.org/ wiki/Neunhundert_neun_und_neunzig_und_noch_etliche_ Almanachs-Lustspiele_durch_den_W%C3%BCrfel) Viktor J. Illmer, Frank Fischer, Mark Schwindt, Jonas Rohe: »999 und noch etliche [mehr]«. Georg Nikolaus Bärmanns »WürfelAlmanach« von 1829 als Web-App. In: DHd2025: »Under Construction«. 3–7March 2025. Book of Abstracts. Bielefeld University. https://doi.org/10.5281/zenodo.14943242 Generative AI for XML-TEI Encoding Exploring the potential of LLMs for Spanish Golden Age Theatre Marco De Cristofaro (2,4); Emanuele Leboffe (1); Daniel Zilio (3) (1) Universitat Autònoma de Barcelona, Spain; (2) Université de Mons; (3:) Università degli studi di Padova; (4) Université de Namur Keywords Spanish Golden Age theater, digital critical editions, Generative Artificial Intelligence, Large Language Models, Digital Humanities Abstract Nowadays, encoding texts in XML-TEI has become a fundamental practice in the field of digital humanities. Its importance lies in creating structured, interoperable representations of texts, enabling their storage in online databases where they can be accessed, consulted, and reused by other scholars, for example, to conduct large-scale quantitative analyses such as au - tomated data extraction or linguistic profiling. However, encoding can be particularly challenging, especially when the texts involved display significant complexity. Spanish Golden Age comedies offer a fitting example. These texts present a layered structure composed of multiple elements — characters’ interventions, verses, stanzas, stage directions — all of LONG PAPER 112 Generative AI for XML-TEI Encoding which require detailed markup. If we also take into account the considerable length of these plays — each averaging around 3,000 verses — encoding can quickly become an arduous and highly mechanical task, threatening the sustainability of research unless effective methods for automating or facilitating the workflow are adopted. Given these complexities, generative artificial intelligence stands out as a promising solution. Its potential for academic research has already been demonstrated in the sciences (Schmidgall et al. 2025) and the humanities (DeRose 2024); moreover, recent developments in AI have influenced how archival materials and literary texts are preserved and accessed (Colavizza et al. 2021), highlighting its effectiveness in supporting the study and conservation of literary archives and cultural heritage more broadly (Carbé 2023). With a view to ensuring research sustainability, this paper proposes an approach to automatic XML-TEI encoding of theatrical texts, using as a case study the approximately 350 Spanish Golden Age plays already encoded and made available by the Biblioteca Digital ARTELOPE23. In the initial phase, we preprocessed the material in two parallel ways: first, we generated simplified XML-TEI files by stripping away non-essential TEI elements; second, we produced plain TXT files containing the same textual content. This dual preparation served different purposes: one portion of the original XML files was used to provide structured references for the model during training and prompt design, while another portion was held back and used as benchmark data to evaluate the accuracy of the final automatic encoding process. To produce these automatic encodings, the plain text files were fed into two different large language models — OpenAI ChatGPT-4 and 23 The corpus of comedies encoded in XML-TEI is available at: https:// gitlab.com/artelope1/ARTELOPE/-/tree/main/XML-TEI%20Playtexts?ref_type=heads (last accessed: 10/07/2025). Anthropic Claude 3.5 Sonnet — using both chat-based interaction and API-based workflows. The chat-based approach revealed a relative independence between the two models, with a very high recognition rate, although consistency tended to decline over the course of the text. The API-based approach, on the other hand, showed greater adherence to the intended encoding. While in both cases we relied on manual verification, for long-term research sustainability, developing a model for automatic validation based on standardized metrics will be crucial. In conclusion, the first results suggest that LLMs hold considerable promise for the automatic encoding of texts characterized by significant length and complexity. Although further refinement is needed, these indications are encouraging and point to generative AI as a potential breakthrough for largescale philological projects. 126 Jinks / TEI Publisher 10 Jinks / TEI Publisher 10 Magdalena Turska e-editiones, Poland Keywords TEI Publisher, editing workflow, digital editions Abstract Due to our modular approach to crafting software, TEI Publisher grew from just two core packages to an entire framework of modules, libraries and repositories, independently released with their own version numbers. Nevertheless, community still likes to think of major new versions of the TEI Publisher. The 10th, anniversary edition, marks a decade since the first Publisher presentation (and our first workshop) at the TEI conference in Lyon. To celebrate the occasion, we would like to present, in the hands-on format, the current state of TEI Publisher’s development, in particular the new Jinks application manager. In practice, Jinks is an edition builder where user can click together an edition and tailor it to her needs in literally minutes, thanks to a new Jinks library of profiles and features. Existing features already cover basic needs of every edition plus quite some very specific requirements, for e.g. parallel views, correspondence or presentation of textual variants. As usual, everyone is invited to bring forward other features to be added to this community-curated collection of building blocks for scholarly publications. WORKSHOP To acknowledge growing community of TEI and Publisher users in Poland, we’d like to offer this workshop primarily in Polish, with “whispered translation” offered for non-Polish speakers. In case of high interest, we could also cover two separate language tracks, Polish and English. 128 JinnTap. A browser-based TEI-XML editor JinnTap A browser-based TEI-XML editor Wolfgang Meier (1,2,4); Lars Windauer (1,2); Joseph Wicentowski (3) (1) e-editiones, Switzerland; (2) Jinntec GmbH, Germany; (3) FSI Office of the Historian, USA; (4) Heidelberger Akademie der Wissenschaften, Germany Keywords TEI, editor, workflow Abstract The formalisation of an edition in an informative structure, e.g. via a mark-up language, that enables its processing is, for many scholars, one of the most challenging aspects of digital editing. This is why several endeavours exist that facilitate textual encoding (such as ediarum24, eLaborate25, FairCopy26, TEI Publisher27 or TextGrid28 to name a few). They aim at assisting humanists in different parts of the editorial pipeline, including transcription, semantic annotation and other aspects. 24 https://www.bbaw.de/bbaw-digital/telota/ forschungsprojekte-und-software/ediarum 25 https://www.elaborate.huygens.knaw.nl/ 26 https://faircopyeditor.com/ 27 https://teipublisher.org 28 https://textgrid.de/ DEMONSTRATION In this demo we will show JinnTap29, a means to edit TEI-XML documents using a browser-based rich text editor, and highlight the differences to existing initiatives. While JinnTap was initially inspired by concrete workflow needs at the Office of the Historian, it is intentionally designed to be broadly applicable. JinnTap preserves the structure of the XML in the editor, thus rooted in the WYSIWYM paradigm, but exploiting the common features of WYSIWYG editors. JinnTap forces the editor to critically reflect on the data model, and create a coherent customization of the TEI schema, tightly fitting the project. While still under development, the default schema already supports: • block level elements like headings, paragraphs, lists, divisions • elements for arbitrarily nested inline formatting like <hi> or <title> • semantic markup for people, organizations, places, terms • analytic elements like <date> • footnotes • inline nodes representing an alternative like <choice>, <abbr>, <expan> • figures, figure descriptions and graphics The schema is fully customizable as long as the distinction between block and inline elements is clear, which we argue is an asset in terms of modelling and not a limitation. 29 https://github.com/JinnElements/jinn-tap 130 LEGOstyle. Building modular, flexible Editions with TEI LEGOstyle. Building modular, flexible Editions with TEI 131 LEGOstyle Building modular, flexible Editions with TEI Thomas Kollatz Academy of Sciences and Literature Mainz, Germany Keywords Digital Edition, Correspondences, CMIF Abstract The ‘Buber-Correspondences Digital’ (BKD) project focuses on the 43,000 preserved letters, postcards and telegrams exchanged between the philosopher of religion, Martin Buber (1878–1965), and his over 7,000 correspondents. Based in Mainz and Frankfurt am Main, this long-term project is funded by the Academies’ Program of the Union of German Academies (AGATE 2025). During the editing process, all correspondence is annotated with the TEI. However, providing a full-text edition is beyond the scope of this project due to the collection’s vast size. Nevertheless, the concept of a modular edition with varying levels of granularity is a viable option. The modular edition BKD has introduced a modular ‘building block system’ into the editorial process. The building blocks, referred to as ‘A’, ‘B’, ‘C’ and ‘U’, respectively represent varying levels of indexing (BKD 2025: 2.2. Erschliessungskategorien – Die modulare Edition). The basic ‘C’ block is included in every edited source (100% of SHORT PAPER correspondence). It contains fundamental information on the sender, recipient and date, i.e. the TEI element <correspDesc>, as well as information on the extent, object type, mode of writing, holding institution and signature. This enables all data to be entered into the web service correspSearch (‘correspSearch – Briefeditionen vernetzen (3.1.0)’ 2025) and to be linked with other correspondence editions. The ‘B’ building block serves to introduce a dynamic element (Jurst-Görlach and Kollatz 2023). It contains a human-readable abstract (<abstract>) as well as lists of entities (persons, organizations, places, events and works) mentioned, and expresses existing relationships between these entities where applicable. The TEI element <relation>, which is used to generate triple-like statements, is employed for this purpose. This particular building block, which is also applied to ‘A’ letters, is used for 65% of all correspondence (Jurst-Görlach 2024). Building block ‘A’ is included in 25% of edited letters; contained in the TEI element <text>, there is a ‘classic’ full-text edition in which the correspondence is transcribed, annotated, and, if necessary, translated. Building block ‘U’ contains correspondence items that have not been preserved, but whose Graphic created by Denise Jurst-Görlach 2025 (CC-BY) 132 LEGOstyle. Building modular, flexible Editions with TEI existence is proven by clear references in existing correspondence or other sources. Conclusion This LegoStyle approach enables flexibility. All information is provided in a standardized ODD-powered scheme, and upgrades can be made at any time by the project team or via reuse by third parties. For example, version 2 of the Correspondence Metadata Interchange Format (‘CMIF V2’ 2023) provides an ideal vocabulary for expressing the BKD approach (Dumont et al., no date): cmif:Record equals building block ‘C’, cmif:Abstract equals building block ‘B’, cmif:Transcription equals building block ‘A’, and cmif:NoTextBase equals building block ‘U’. The generated data is made available in the BKD repository from the outset (‘Repositorium des Akademievorhabens BuberKorrespondenzen Digital’ 2022). References AGATE. 2025. „Buber-Korrespondenzen Digital. Das Dialogische Prinzip in Martin Bubers Gelehrtenund Intellektuellennetzwerken im 20. Jahrhundert“. A European Gateway for the Academies of Sciences and Humanities. https://agate.academy/id/PR768. BKD. 2025. „Buber-Korrespondenzen Digital. Das Dialogische Prinzip in Martin Bubers Gelehrtenund Intellektuellennetzwerken im 20. Jahrhundert (BKD). Editionsrichtlinien und Schema“. https://adwmainz.pages. gitlab.rlp.net/digicademy/bkd/correspondences/schema/ tei_bkd.html. CMIF V2. 2023. https://github.com/TEI-Correspondence-SIG/ CMIF/blob/main/proposals/CMIFv2_proposal.md. correspSearch – Briefeditionen vernetzen (3.1.0). 2025. Edited by Stefan Dumont, Sascha Grabsch, Jonas Müller-Laackman, Ruth Sander and Steven Sobkowski. Berlin-Brandenburgische Akademie der Wissenschaften 2025. https://correspSearch.net. Dumont, Stefan, Ingo Börner, Dominik Leipold, Martin Anton Müller, Jonas Müller-Laackman, Klaus Rettinghaus, Gerlinde Schneider, Torsten Schrade, Peter Stadler, and Jakub Šimek. o.J. „DRAFT: Correspondence Metadata Interchange Format (CMIF) vocabulary“. TEI Correspondence SIG. https://lod. academy/cmif/vocab/terms/. Jurst-Görlach, Denise. 2024. „Relationen dokumentieren in TEI – Kommentieren im Semantic Web“. In KONDE Weißbuch. Edited by Selina Galka and Helmut W. Klug in collaboration with Susanne Höfer in the project „Enlarging ‚Weißbuch Digitale Edition‘“. https://www.digitale-edition.at/o:konde.250. Jurst-Görlach, Denise, and Thomas Kollatz. 2023. „Text Encoding without //text. The use of //abstract as means to avoid the one-dimensionality of ego-networks in ›BuberCorrespondences Digital‹ project“. Paderborn. https://gitlab.rlp.net/adwmainz/digicademy/bkd/bkd-presentations/ teimec. Repositorium des Akademievorhabens Buber-Korrespondenzen Digital 2022. https://gitlab.rlp.net/adwmainz/digicademy/ bkd/correspondences. 134 Linking Your *-ographies… Linking Your *-ographies Developing project-specific TEI Authority File Lookups for LEAF-Writer James Cummings (1); Luciano Frizzera (2); Diane Jakacki (3); Susan Brown (4); Mihaela Ilovan (5); Kelsey Rubin-Detlev (6) (1) Newcastle University, UK; (2) University of Waterloo, Canada; (3) Bucknell University, USA; (4) University of Guelph, Canada; (5) University of Alberta, Canada; (6) University of Southern California, USA Keywords TEI XML, Web Editor, Linked Open Data, TEI Authority Files, Named Entities Abstract LEAF-Writer is a popular, free, web-based, semantic editor for Text Encoding Initiative (TEI) and Linked Open Data (LOD) annotation. While LEAF-Writer is available in a standalone version that anyone can use (LEAF-Writer Commons), it is also a crucial component of the larger LEAF-VRE environment. LEAF-Writer Commons runs entirely in your browser and files are saved to your own GitHub repositories (or downloaded locally), with no data stored on LEAF servers. LEAF-Writer provides an easy text-first editing interface that encourages the encoder to focus on adding semantic markup, scholarly notes, and identifying named entities. You can choose to edit in the tags-off view, show tags, or edit the SHORT PAPER underlying TEI XML. LEAF-Writer includes schema-constrained context-sensitive tagging and validation using out-of-the-box popular TEI customizations, or use your own custom TEI project schemas (with your own CSS). One of LEAF-Writer’s most important features is its built-in support for named entity linking. This enables tagging names of people, places, organisations, or works and associating these (with both TEI markup and LOD) to recognised authorities (such as VIAF, Wikidata, DBPedia, Getty, Geonames, GND and LINCS). LEAF-Writer can also generate LOD Annotations from already tagged XML references. Recently, the LEAF team, in collaboration with the CatCor (Correspondence of Catherine the Great) project has introduced functionality for adding web-accessible project-specific TEI Authority files for particular entity types. (e.g. a TEI ‘personography’ file containing a <listPerson> for person entities, <listPlace> for place entities, etc.). This exciting new feature is for many TEI projects that want to do look-ups using their own project-specific ‘*-ography’ authority files. LEAF-Writer was introduced at TEI2022 and we have run workshops and presented new features to the TEI community in the years since. For this conference, we will concentrate on these ‘new territories’ of project-specific TEI Authority Files for named entity lookups. 136 Managing lexical complexity… Managing lexical complexity… 137 Managing lexical complexity ODD chaining for Arabic dialect dictionaries Veronika Engler (1); Karlheinz Mörth (1); Stephan Procházka (2); Michaela Rausch-Supola (1); Daniel Schopper (1) (1) Austrian Academy of Sciences, Austria; (2) University of Vienna, Austria Keywords lexicography, corpora, linguistics, data modelling, ODD chaining Abstract The Vienna Corpus of Arabic Varieties (VICAV) is a language documentation platform hosting a varied collection of digital language resources. From its beginnings, it has adopted a text-oriented approach that has been grounded in the application of the Guidelines of the Text Encoding Initiative (TEI). The VICAV infrastructure is meant to ensure consistent encoding across projects as well as sustainable creation and publishing workflows. In addition to a bibliography of research literature, typologically similar to what can be found in the World Atlas of Language Structures or the Database of Arabic Dialects, VICAV offers several types of data: concise descriptions of linguistic varieties, structured lists of linguistic features, sample texts, corpora of unmonitored speech, and dictionaries (Procházka et al. 2015). LONG PAPER The published VICAV dictionaries so far cover five linguistic varieties: Baghdad, Cairo, Damascus, Tunis and Modern Standard Arabic. These dictionaries are all comparatively small, none of them containing more than 8000 entries, and constitute lexical databases with structured lexicographic information (Moerth and Schopper 2021). They are all provided with English translation equivalents (some also include German, French or Spanish translations) and share common encoding conventions. A recent addition in this series is the Shawi Dictionary, which will go online in late 2025. This is the first VICAV dictionary natively encoded in TEI Lex-0, a relatively new initiative within the TEI community, intended as a baseline encoding for lexicographic data (Tasovac et al. 2018) which meanwhile has been adopted by more dictionary projects (Salgado et al. 2019). Stemming from different projects, all VICAV dictionaries have subtle differences in their encoding requirements. Moreover, there are several other comparable lexicographic resources being developed at the ACDH-CH. To be able to document the differences between those dictionaries and yet keep them compatible, we use the method of ODD chaining, which is the process of deriving ODDs from one another (Pernes et al. 2017). In our case, we proceeded from the TEI Lex-0 ODD as the primary source and modified it to create the ACDH-CH generic-dict-schema, which serves as the baseline ODD for various VICAV and other dictionaries worked on at the ACDH-CH. Since, for example, TEI Lex-0 does not make prescribe the macrostructure of a dictionary, it is in this schema that we define that in our dictionaries examples are not embedded within entries but kept in a separate <div> in order to allow them to be re-used in different contexts. In a third step, the SHAWI dictionary ODD is derived from the generic-dict-schema and provides specifications adapted to the needs of the SHAWI project. Unlike in other dictionaries, the lexical profiles of tribes and their geographic context play an important role. To incorporate this information, we extend generic-dict-schema to allow typed <name> 138 Managing lexical complexity… elements within <usg> – a construct which is not needed in the other dictionaries. In our paper, we will present the dictionaries involved and our general mechanism next to discussing pros (modularization; specificity) and cons (complexity both in terms of processing overhead and modelling) of having selected this relatively complex route. References Dryer, Matthew S. & Haspelmath, Martin (eds.) 2013. WALS Online (v2020.4) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.13950591 (Available online at https://wals.info ) Magidow, Alexander. 2015. Database of Arabic Dialects. (Available online at http://database-of-arabic-dialects.org/ ) Moerth, K. and D. Schopper (2021) “VICAV 3.0: Zooming in on Lexical Resources”, in C. Katsikadeli, M. Sellner and M. Gassner (eds.) Digital Lexis, and Beyond. Selected Papers from the Workshop „Digital Lexis, and Beyond” 45th Austrian Linguistics Conference 2019. Wien: Verlag der ÖAW. Pernes, S., L. Romary and K. Warburton (2017) “TBX in ODD: Schema-agnostic specification and documentation for TermBase eXchange”, in F. Frontini, L. Grcic, Š. Vintar and F. Khan (eds.) Proceedings of the Language, Ontology, Terminology and Knowledge Structures Workshop (LOTKS 2017). Procházka, S. and K. Moerth (2015) “The Vienna Corpus of Arabic Varieties: building a digital research environment for Arabic dialects”, in M. Al-Hamad, R. Ahmed and H. Aloui (eds.) Lisan AlArab: Studies in Contemporary Arabic Dialects, Proceedings of the 10th International Conference of AIDA, Qatar University 2013. Vienna: LIT Verlag: 209-218. Salgado, A., R. Costa, T. Tasovac and A. Simões (2019) “TEI Lex0 In Action: Improving the Encoding of the Dictionary of the Academia das Ciências de Lisboa”, in Electronic lexicography in the 21st century. Proceedings of the Electronic lexicography in the 21st century (eLex 2019): 417-433. https://elex.link/ elex2019/wp-content/uploads/2019/09/eLex_2019_23.pdf. Tasovac, T., L. Romary, P. Banski, J. Bowers, J. de Does, K. Depuydt, T. Erjavec, A. Geyken, A. Herold, V. Hildenbrandt, M. Khemakhem, B. Lehečka, S. Petrović, A. Salgado and A. Witt (2018) TEI Lex-0: A baseline encoding for lexicographic data. Version 0.9.3. DARIAH Working Group on Lexical Resources. https://dariah-eric.github.io/lexicalresources/pages/TEILex0/ TEILex0.html. [Accessed 2025-03-27] TEI Consortium, eds. Guidelines for Electronic Text Encoding and Interchange. [last modified 24.01.2025]. (Available online at https://www.tei-c.org/P5/) 140 Multimodality and Minimal Publishing… Multimodality and Minimal Publishing… 141 Multimodality and Minimal Publishing TEI, MEI, and more in 19th-Century Music Treatises Torsten Roeder; Jana Klinger; Felicitas Stickler; Corinna Keupp; Fabian Moss Julius-Maximilians-Universität Würzburg, Germany Keywords Digital Scholarly Edition, Text and Music, Minimal Publishing, Multi-Modality Abstract Introduction While presentation methods for the digital edition of either text or music notation can both rely on established tools for TEI or MEI presentation, digital editions that constantly alternate between modalities—e.g., in the context of music theory treatises, textbooks, music journals, composers’ correspondence, and editions of sheet music with accompanying peritexts—face significant challenges when combining the related technologies. This paper builds on a bachelor’s thesis dealing with source materials from the project Digitizing the Dualism Debate (cf. Moss/Bavaud et al. 2021; Moss/Köster et al. 2021; Moss/Nápoles López et al. 2022). It also connects to the recently reactivated TEI Music SIG, which aims to harmonize and integrate music encoding with the practices of the Text Encoding Initiative LONG PAPER (TEI), focusing particularly on the transitional areas between the two technological stacks. The overarching goal is to offer one or more accessible and sustainable technological solutions for publishing digital editions that include music notation. This approach is intended to support projects without institutional infrastructure for digital editions. The integration of text and music notation initially appears to be a topographical issue of text presentation. From a text en - coding perspective, the problem can be addressed by integrating graphical representations of music notation as digital images. However, from a music encoding perspective, this approach is insufficient, as it prevents the formulation of relationships between text and music notation within the data (cf. Roeder/Moss/ Köster 2023). Therefore, the objective is to model and encode both text and music with comparable structural and semantic depth, and to host it within a long-term sustainable environment. While the presentation could integrate scanned or vectorized music notation graphics alongside embedded MEI data, the preferred approach is to generate the presentation directly from the data without intermediate formats (while providing a fallback solution with embedded graphics if necessary). To illustrate this approach, we propose using two tools closely associated with TEI and MEI: Verovio for rendering music notation and CETEIcean, a lightweight tool for TEI processing. Both operate on JavaScript and require only a standard web server, as TEI and MEI are both processed client-wise. 1. Representation: Organizing and Modeling the Data Due to the different encoding standards, it is necessary to decide how to organize text and music notation data at the file level. To avoid potential namespace conflicts between frameworks, separating them into distinct files has proven effective. This does not necessarily mean generating one TEI file and numerous MEI files; MEI data could also be consolidated into a single file and referenced section by section using IDs. 142 Multimodality and Minimal Publishing… Multimodality and Minimal Publishing… 143 For example, a treatise with alternating text and musical examples could use a primary TEI file for the text and a single MEI file containing all musical excerpts, each tagged with unique IDs for cross-referencing. This approach reduces complexity while maintaining semantic depth. 2. Presentation: Rendering the Data Depending on the chosen data organization, the rendering process must be arranged for the browser. With the selected combination of CETEIcean and Verovio, HTML and SVG code are generated live using JavaScript. This requires precise timing: a music notation section can only be rendered by Verovio once the corresponding HTML element has been generated by CETEIcean. To achieve this synchronization, the frameworks are orchestrated via monitoring. When CETEIcean processes an XML element referencing a musical excerpt, it temporarily hands over control to Verovio to render the notation dynamically. This interplay ensures a seamless user experience without relying on pre-generated images or static files. 3. Stabilization: Hosting Environment and Citable Archiving Dynamic site digital editions are often precarious resources, as they often depend on complex server environments. Even standard solutions (e.g. with TEI Publisher) require regular maintenance, which projects and their institutions are not able to guarantee in long terms. For rendering components such as CETEIcean and Verovio, however, simple hosting solutions for static sites like GitHub Pages or any another web server are sufficient (cf. Cayless/Viglianti 2018). GitHub offers advantages such as temporal organization of data and code into release sequences. Additionally, GitHub can be linked to Zenodo, enabling the repository and its individual releases to obtain DOIs and ensuring that each release is archived independently of GitHub on Zenodo. This combination allows projects to achieve both low-cost deployment and academic curation, compliant with the endings principles (cf. The Endings Project 2023). For instance, an edition hosted on GitHub Pages could be continuously updated, while its key versions remain permanently accessible via Zenodo. Outlook The Würzburg project DigiMusTh: Aufbau einer offenen digitalen Sammlung historischer musiktheoretischer Texte aus dem deutschsprachigen Raum anhand von Beispielen aus dem 19. Jahrhundert (“Development of an open digital collection of historical music theory texts from the German-speaking world based on examples from the 19th century”) will adopt this new presentation model in 2025 using the described approach. Simultaneously, this approach will serve as an example for the TEI Music SIG, showcasing its potential for other interdisciplinary projects. The next step involves modeling the dense relationships between text and music notation and integrating these into the presentation. It will also be discussed and presented how to include further rendition methods for other modalities, such as mathematical formulas in MathML or diagrams in SVG. References Cayless, H., Viglianti, R. (2018). CETEIcean: TEI in the Browser. In Proceedings of Balisage: The Markup Conference 2018. Balisage Series on Markup Technologies, vol. 21. https://doi. org/10.4242/BalisageVol21.Cayless01. CETEIcean (2024). Release 1.9.0 (Jan 10), https://teic.github.io/ CETEIcean. The Endings Project (2023). Endings Principles for Digital Longevity, Version 2.2.1, 2023-03-03. https://endings.uvic. ca/principles.html. Moss, F. C., Bavaud, F., Métrailler, C., Femminis, M., Köster, M. (2021). Digitizing the Dualism Debate. https://dcmlab.github.io/ddd/. Navigating and Processing Data from the TEI with XPath and XSLT 145 Moss, F. C., Köster, M., Femminis, M., Métrailler, C., & Bavaud, F. (2021). Digitizing a 19th-Century Music Theory Debate for Computational Analysis. In M. Ehrmann, F. Karsdorp, M. Wevers, T. L. Andrews, M. Burghardt, M. Kestemont, E. Manjavacas, M. Piotrowski, & J. van Zundert (Eds.), CHR 2021: Computational Humanities Research Conference, November 17–19, 2021, Amsterdam, The Netherlands (pp. 159–170). CEUR. http://ceurws.org/Vol-2989/short_paper31.pdf. Moss, F. C., Nápoles López, N., Köster, M., & Rizo, D. (2022). Challenging sources: A new dataset for OMR of diverse 19th-century music theory examples. In J. Calvo-Zaragoza, A. Pacha, & E. Shatri (Eds.), Proceedings of the 4th International Workshop on Reading Music Systems (WoRMS 2022) (pp. 4–8). https://sites.google.com/view/worms2022/proceedings. Roeder, T., Moss, F. C., Köster, M. (2023). Music-Text Interlinking as a Challenge for Digital Encodings of Music-Theoretical Writings, TEI/MEC 2023. https://teimec2023.uni-paderborn. de/contributions/124.html. Verovio (2024). Version 4.4 (Dec 17). https://www.verovio.org. Navigating and Processing Data from the TEI with XPath and XSLT Elisa Beshero-Bondar (1); Martina Scholger (2); Patricia O’Connor (3) (1) Penn State Erie, USA; (2) University of Graz, Austria; (3) Maynooth University, Ireland Keywords XPath, XSLT, pull processing, data visualization, internationalization Abstract TEI markup provides structures that are particularly useful for processing data beyond what we can do with so-called “plain text”. Our workshop teaches the pull-processing of data from XML/ TEI with simple, reusable XSLT templates to represent in simple TSV/CSV, HTML tables/charts, and (if time!) simple SVG graphics. Knowing how to locate and explore data in your encoding can help to learn how to work with TEI and XML generally. This half-day workshop is designed for people who have some experience with TEI and seek to learn how to work with XML markup for analysis and research. Participants will gain a working, practical knowledge of the query language XPath and the transformation language XSLT, and learn how these can help to reduce reliance on software, packages and plugins that may become obsolete without warning. Further, XSLT’s functional programming can serve as a way of articulating research questions around a document data model expressed in XML. WORKSHOP 158 ODD-API. A REST Interface for Programmatic Use… ODD-API. A REST Interface for Programmatic Use… 159 will be further expanded with version 2 of the API, making the ODD-API usable as a central hub for ODD schema definitions in general, creating various new application possibilities. The API can thus serve as a foundation for developing tools to compare ODD schemas (or different versions of a schema), to edit ODD documents, to visualize dependencies between ODD schemas, for learning platforms to teach ODD knowledge, or dynamic documentation systems to complement existing static websites. Implementation As an initial implementation for the ODD-API scheme, we relied on the eXist database and XQuery. Various ODD schema definitions (e.g., teip5subset.xml) are stored in defined database collections by type and version (e.g., /db/apps/odd-api/data/tei/v4.9.0); additional ODD schema definitions may be loaded by dedicated endpoints. Currently, the following endpoints for GET requests are implemented: • /v2/schemas: returns a list of all available schemas in JSON:API format • /v2/{sche ma}: returns a list of all versions of a schema in JSON:API format • /v2/{sche ma}/{version}: returns the version of a schema as “Compiled TEI ODD” in XML format • /v2/{schema}/{version}/elements: returns a list of all elements in JSON:API format • /v2/{schema}/{version}/attributes: returns a list of all attributes in JSON: API format • /v2/{sche ma}/{version}/classes: returns a list of all classes in JSON:API format • /v2/{schema}/{version}/modules: returns a list of all modules in JSON:API format The response for the endpoint /v2/tei (on a local test instance at http://localhost:8080) in JSON:API format looks like this: { “data”: [ { “type”: “versions”, “id”: “version_tei_4.6.0”, “attributes”: { “version”: “4.6.0” }, “links”: { “self”: “http://localhost:8080/v2/tei/4.6.0” } }, { “type”: “versions”, “id”: “version_tei_4.8.1”, “attributes”: { “version”: “4.8.1” }, “links”: { “self”: “http://localhost:8080/v2/tei/4.8.1” } } ], “links”: { “self”: “http://localhost:8080/v2/tei” } } According to the JSON:API specification, each object must contain a type attribute and an id attribute. The latter must be unique within the API, so in this example, the version number is prefixed with the type and the schema name. The pure version number is returned in the attributes object. Application Scenarios and Outlook A concrete current application scenario for the ODD-API is its use as an information resource for the ODD editor Roma. Roma is to be further developed to allow editing of non-TEI schemas and requires canonical addresses of ODD resources as starting points for customizations. Here, the ODD-API can serve as a central point to provide this information. To enable this in other contexts as well, only the URL of the ODD-API (analogous 160 ODD-API. A REST Interface for Programmatic Use… Parallel Editing in TEI… 161 to the URL of the TEIGarage) would need to be specified as a parameter in the Roma code. Private Roma instances could then address any other ODD-APIs with any ODD schema definitions without having to adapt the source code of Roma or the ODD-API. Beyond that, it would also be possible to open the ODD-API for PUT requests (and potentially also POST and PATCH) and thus also allow adding and editing ODD schemas. This would enable even stronger dynamization of the ODD-API and could be used, for example, in automated pipelines. However, this would then require appropriate authentication and authorization as well as guidelines for handling additions by third parties. Finally, it would also be possible to develop an ODD package manager, i.e., an “npm for ODD schemas.” This package manager could have an integrating effect for the ODD community by highlighting the diversity and possibilities of ODD schemas on the one hand, while promoting standardization on the other. The API defined here would provide the underlying engine for such a hypothetical package manager. References Katz, Yehuda, Dan Gebhardt, Gabe Sullice, and Jeldrik Hanschke. 2022. “JSON:API Specification.” https://jsonapi.org/ OpenAPI Initiative. 2024. “OpenAPI Specification v3.1.1.” https:// spec.openapis.org/oas/v3.1.1 Viglianti, Raffaele. 2019. “One Document Does-it-all (ODD): a language for documentation, schema generation, and customization from the Text Encoding Initiative.” Presented at Symposium on Markup Vocabulary Customization, Washington, DC, July 29, 2019. In Proceedings of the Symposium on Markup Vocabulary Customization. Balisage Series on Markup Technologies, vol. 24 (2019). https://doi. org/10.4242/BalisageVol24.Viglianti01. Parallel Editing in TEI The Case of Regestra 1561 Łukasz Poznański; Bogumił Szady Humanities Research Data Lab, The John Paul II Catholic University of Lublin (KUL), Poland Keywords TEI, parallel edition, historical documents, linked data, TEI Publisher Abstract This paper presents Regestra 1561, a parallel digital edition of two manuscript copies of a clerical tax register from the diocese of Poznań. Both documents, preserved in the Central Archives of Historical Records in Warsaw, were compiled following the 1561 provincial synod. They record contributions from local clergy, including names, ecclesiastical benefices, places, and amounts of taxes owed to the Crown. As part of the Regestra 1561 project, both a traditional fulltext print edition and a digital edition were prepared. The digital edition was designed to enable a side-by-side comparison of both witnesses. Encoded in TEI-XML, it captures the full text of each manuscript and aligns corresponding sections, allowing users to explore both similarities and differences in content and structure. Creating such an edition posed challenges: the manuscripts diverge in layout and pagination, requiring a parallel markup SHORT PAPER Processable and Computable Media in TEI… 163 system that preserves each witness’s structure while linking related segments. The edition also includes a TEI-encoded critical apparatus and detailed references to people and places. The edition is presented in TEI Publisher with customized views supporting synchronized navigation of both manuscripts. Thanks to IIIF integration, the texts appear alongside digitized folios. Differences between the versions are automatically highlighted. The edition also links to a structured database hosted on wiki.kul.pl, which offers semantic context for named entities. To enable searching and filtering via TEI Publisher, the contents of this external database were exported and integrated into the TEI source files. This allows users to query and explore the edition using enriched metadata based on normalized person and place records. The project illustrates how TEI-based parallel editing, combined with linked data and visual tools, can offer an interactive presentation of early modern administrative documents. Processable and Computable Media in TEI Usecases and Strategies Torsten Roeder; Tomash Shtohryn; Yannik W. Herbst University of Würzburg, Germany Keywords paper forms, e-literature, digital correspondence, program code, punchcards Abstract Beyond Readability? In everyday usage, “text” typically refers to physical or electronic inscriptions — material or digital — that are intended for human reading. Our notion of textual heritage predominantly encompasses classical forms such as literature, correspondence, journals, and newspapers, all seemingly bound to natural language. The TEI Guidelines offer extensive support for encoding such textual complexity, accommodating a wide range of human-created text types. Although such texts are, in principle, amenable to processing by computational rulesets or human interpretation, this kind of processing is not usually the primary focus. This prompts a set of critical questions: Is “text” — and by extension, the TEI — necessarily bound to human natural POSTER 164 Processable and Computable Media in TEI… Processable and Computable Media in TEI… 165 language? Conversely, are there classes of textual heritage that require mechanical or rule-based processing either before, during, or after human or machine reading? If so, how can these be encoded within the existing TEI framework, and what expansions might be necessary? Furthermore, how can the rulesets for such processing themselves become integral to the preservation of cultural heritage? Examples The questions raised may appear broad, if not esoteric, thus it is appropriate to illustrate them with concrete examples. Consider a multiple-choice examination sheet completed by a student and graded by a teacher. The form itself encodes an implicit ruleset that governs how the examination is to be completed and assessed — thus undergoing multiple stages of processing to produce a graded result. Another example stems from early digital literature: HyperCard-based works require specific computational environments (e.g., pre-OS X Macintosh systems) for proper functionality. The user’s interaction with these works is defined by both the hardware and software, raising the question of whether essential aspects of historical digital interfaces can be preserved through TEI encoding (cf. Ensslin 2007 for a more general discussion on early hypertext fiction). Similarly, early digital periodicals such as “diskmags” — early software magazines distributed on floppy disk — often employed custom character sets and graphical conventions that defy standard extraction and representation methods. Here, knowledge of the original processing mechanisms is essential for reconstructing the intended presentation of text, images, and audio content (cf. Shtohryn 2025). Emails, a contemporary form of correspondence, further complicate notions of text (cf. Beshero-Bondar and Bauman 2024). Beyond metadata such as sender and recipient information, the software environment critically influences the creation, display, and transmission of emails. Variations between client applications and server-side modifications introduce layers of textual genetics and performance dynamics that are relevant for scholarly documentation. Finally, historical program code published in print media (“paperware”) exemplifies the hybrid nature of text and process (cf. Höltgen 2014). Such code can be “read” and executed by both humans and machines. Punchcards, featuring both computer-readable perforations and human-readable printed text, reinforce the idea that digitality often straddles material and electronic forms (cf. Feichtinger 2023). Perspectives These examples collectively demonstrate that distinctions between “digital” and “material” documents are secondary when considering a document’s intentional processability. While this poster centers primarily on born-digital heritage, it recognizes the blurred boundaries between digital and material media. Recent TEI developments — such as the introduction of the <post> element by the special interest group for “ComputerMediated Communication” in 2025 (cf. TEI 2025) — illustrate growing recognition of these evolving forms. Given the increasing volume of born-digital cultural heritage and its vulnerability to hardware decay, software obsolescence, corporate restrictions, and political censorship, it is urgent to develop strategies for preserving processable and computable forms of text. This poster will present the aforementioned examples to initiate a discussion on extending TEI’s capabilities in this area. We aim to establish a new TEI Special Interest Group dedicated to processable and computable media — recognizing that the tra - ditional notion of “text” may not be sufficiently inclusive in this context. Our initial focus will be on utilizing existing guidelines, proposing semantic extensions where necessary, and, where appropriate, developing new attributes or elements. Reconstructing the Experience of a Historian from the Meiji… 167 References Beshero-Bondar, Elisa; Bauman, Syd (2024): Can we apply the new CMC chapter to the TEI Listserv Archives? An experiment with TEI for Correspondence and Computer-Mediated Communication, TEI Conference 2024. https://zenodo.org/ records/13957666 Ensslin, Astrid (2007): Canonizing Hypertext. Explorations and Constructions. Feichtinger, Moritz (2023): Annotation, Simulation und Analyse eines historischen Datenbanksystems. In: Burghardt, M. & Weiß, C. (eds.): Lecture Notes in Informatics (LNI), Gesellschaft für Informatik. Höltgen, Stefan (2014): Humanities of the Digital. Philologische Perspektiven auf Source Codes als Beitrag einer computerarchäologischen Knowledge Preservation. In: Bartelmus, M./Nebrig, A. (eds.): Digitale Schriftlichkeit. Programmieren, Prozessieren und Codieren von Schrift, p. 207–229. Shtohryn, Tomash (2025): Semi-automatisierte Erzeugung eines Textkorpus von deutschsprachigen Diskettenmagazinen für das Heimcomputersystem Commodore 64 im TEIFormat, Universität Würzburg. https://doi.org/10.25972/ OPUS-40282 TEI (2025): Computer-mediated Communication. In: TEI: Guidelines for Electronic Text Encoding and Interchange. P5 Version 4.9.0, 24th January. https://tei-c.org/release/doc/ tei-p5-doc/en/html/CMC.html Reconstructing the Experience of a Historian from the Meiji (明治) Period Attempts to Encode Kengo Murakawa (村川堅固)’s Diary in TEI Jun Ogawa (1); Yuko Fukuyama (2); Shozo Matsutani (3) (1) The University of Tokyo, Japan; (2) Waseda University, Japan; (3) Japan Society for the Promotion of Science, Japan Keywords East Asian materials, diary, writing mode, historiography Abstract In this paper, we discuss the Murakawa Collection, a compilation of documents related to Kengo Murakawa (1875–1946), a significant figure in Japanese academia, particularly in the field of Western ancient history [1]. It encompasses various documents such as postcards, lecture notes, and household accounts. Here, we specifically focus on his diary as a case study for TEI encoding because it enables us to gain insights into his live experiences and thoughts. SHORT PAPER 168 Reconstructing the Experience of a Historian from the Meiji… Reconstructing the Experience of a Historian from the Meiji… 169 As TEI encoding methods for diary documents have already been proposed in previous studies like The Harry Watkins Diary [2] and Shibusawa Eiichi Diary [3], we decide to adopt them to encode the basic structure of our texts. Each entry, typically corresponding to a single date (see Fig. 1), is structured using <div> with @type=’entry’, and physical page breaks are marked with <pb/>. Header information such as date, location, and weather, often found at the start of each entry, is encoded with <dateline>, while entities like personal names and places are also tagged with <persName>, <placeName>, etc. as shown in Fig. 2. In addition to structure and entities, we link text to corresponding images at the page level by adding an @facs value to each <pb/>, referencing the <graphic> element in the <facsimile> block. Our encoding also addresses text orientation, a unique feature of our diary which includes both vertical and horizontal text, often requiring the book to be rotated for proper reading, preserved through @style for each <pb/> [4]. Our project tackles significant issues in historical Japanese handwritten sources, such as text orientation and the integration of Japanese and Western languages. We apply established diary encoding methods to our specific sources and are developing a diary viewer (Fig. 3) using structured TEI texts, advancing digital preservation and analysis of our materials. Acknowledgement We are profoundly grateful to Natsuko Murakawa for kindly granting permission to view and digitally preserve the Murakawa family materials discussed in this paper. We also deeply appreciate the detailed insights she provided regarding the contents of these materials. Fig. 1: Basic structure of a single diary entry WeatherDate Entry Fig. 2: Markup example of a selected entry from the diary 170 Reconstructing the Experience of a Historian from the Meiji… Save it on a Shoe String… 171 References [1] www.l.u-tokyo.ac.jp. (n.d.).The History of the Faculty of Letters - Graduate School of Humanities and Sociology / Faculty of Letters, University of Tokyo. [online] Available at: https://www.l.u-tokyo.ac.jp/eng/history/history.html [Accessed 11 Jul. 2025]. [2] Umich.edu. (2018).Index | The Harry Watkins Diary: Digital Edition | University of Michigan Library Digital Collections. [online] Available at: https://quod.lib.umich.edu/h/hwatkins/ [Accessed 11 Jul. 2025]. [3] 渋沢栄一ダイアリー. (2025).渋沢栄一ダイアリー. [online] Available at: https://shibusawa-dlab.github.io/app1/ [Accessed 11 Jul. 2025]. [4] Tei-c.org. (2025).5 Characters, Glyphs, and Writing Modes - The TEI Guidelines. [online] Available at: https://www.tei-c. org/release/doc/tei-p5-doc/ja/html/WD.html#WDWMEG [Accessed 11 Jul. 2025]. Fig. 3: TEI-based viewer for Murakawa’s diary Save it on a Shoe String Experiences with Project Endings Constance Crompton University of Ottawa, Canada Keywords sustainable web development, metadata, search Abstract The Humanities offer superpowers to the public: the ability to understand the past, avoid its mistakes, and make evidence-backed decisions about the future (Nussbaum). When used in online knowledge creation and dissemination, Humanities superpowers have the might to stave off an intellectual dark age, provided that we prevent a digital dark age characterized by the disappearance of Humanities data from the Web ecosystems (Barats; Carlin et al 2023a; Crompton 2023; Maron et al.). We know the Web ecosystem is fragile as we have experienced this die-off on the Web before: from its earliest days, the Web was a way to amplify marginalized voices from the past. Indeed, throughout the 1990s, many “scholars invested in early work on race [and class and gender in] the digital humanities insisted on building editions and digital texts as an activist intervention in the closed canon” (Earhart, 317). Many of the earliest digital project content has disappeared from the Web when it in formats that could not be sustained by home institutions (Goddard and Seeman; Jenstad, Holmes SHORT PAPER Standardisation and innovation. The role of TEI… 173 and Huculak). This has resulted in the loss in particular of online Humanities scholarship, or in partial snapshots of projects captured by third-party services, like the Internet Archive. There is much room to better serve and preserve Humanities projects and to continue to share their vital knowledge in the way we do Humanities monographs without increasing the cost and technological burden to academic libraries and other stakeholders. This short paper introduces the Lesbian and Gay Liberation in Canada project’s experience with the principles and code based developed by the Endings Project at the Humanities Computing Media Centre at the University of Victoria (“Endings”). Lead by the tech lead who first used Endings beyond the original Endings team, we outline the benefit of switching to a static-only version of the site, and the benefits of this approach in the current North American financial and political context. Standardisation and innovation The role of TEI in improving innovative potential in digital scholarly editions of correspondences Agnieszka Szulińska Polish Academy of Sciences, Poland Keywords digital scholarly editions, scholarly communication, correspondence, letter Abstract In this talk, I want to present the first phase of a broader project exploring digital scholarly editions as innovations in scholarly communication and, more specifically, the role of the TEI standard in enhancing their innovation potential. Innovation, understood as “novel approaches, tools, or workflows that enhance the creation, dissemination, and evaluation of scholarship” (Maryl, Wnuk, Gouzi and Umerle 2024), could become an important point influencing the perception of scholarly digital editing in Poland. Digital editions have been created in more Polish scholarly insitutions relatively recently, and many scholars are still unsure whether it is worth publishing the SHORT PAPER Starting from Showing, Moving toward Sharing… 175 results of their work in digital form. Analyses of the innovative potential of digital scholarly editions can help editorial teams decide on the medium for the research effects presentation. Using the example of selected digital correspondence editions, which are one of the most numerous groups of projects of this type, I intend to trace how the TEI standard affects areas (Maryl et al 2021) such as edition authors and recipients, data and software, and dissemination and evaluation. Among the editions analyzed were such as Maria Dabrowska – Anna Kowalska. Letters 1940–1965 published on the TEI Panorama platform or selected digital correspondences published on the TEI Publisher Registry such as Karl Barth Gesamtausgabe. References Maria Dabrowska – Anna Kowalska. Letters1940–1965, https:// tei.nplp.pl/categories/korespondencja-marii-dabrowskiej-i-anny-kowalskiej Maryl, Maciej, Marta Błaszczyńska, Agnieszka Szulińska, et al. 2021. ‘OPERAS-P Deliverable D6.5: Report on the Future of Scholarly Writing in SSH’. https://doi.org/10.5281/ zenodo.4922512 Maryl, Maciej, Magdalena Wnuk, Françoise Gouzi, and Tomasz Umerle. 2024. ‘Guidelines on Publishing and Evaluating Innovative Outputs in Social Sciences and Humanities’, November. https://zenodo.org/records/14221728 TEI Publisher Registry, https://www.e-editiones.org/map/ Starting from Showing, Moving toward Sharing Broadening TEI’s Reach Akihiro Kameda (1); Kohei Ishii (2); Satsuki Inoue (2); Naoki Kokaze (2) (1) National Institutes for the Humanities, Japan; (2) Chiba University, Japan Keywords TEI, HTML, research data management Abstract This study proposes Tidy Text Principles and workflows for structuring textual data. This approach begins with a meaning-oriented representation (“showing”) aligned with specific research objectives and proceeds to full structuring and sharing. Two datasets are presented as case studies. The first involves political documents and parliamentary records, using lightweight TEI XML aligned with the Parla-CLARIN schema, supplemented by HTML and JSON for visualization and annotation. The second addresses Wittgenstein’s Tractatus Logico-Philosophicus, where an index dataset was restructured in JSON, then converted to TEI, focusing on mathematical content and multilingual term relationships. In both cases, preprocessing—transforming unstructured or inconsistently POSTER Streamlining TEI Workflows… 177 structured texts into tidy forms—posed the main challenge, whereas TEI conversion itself proved straightforward once a tidy intermediate format had been established. “Tidy text” is defined not by strict data normalization, but as a pragmatic, research-driven organization: units of interest are clearly segmented, minimally nested, and addressable. While TEI is often used as a starting point in textual studies, we suggest that workflows beginning elsewhere—due to practical or technical needs—can still reach TEI as a meaningful endpoint, particularly when tidy structuring is applied. The principles proposed herein remain preliminary and are intended to invite feedback during the presentation, with the aim of refining them collaboratively according to diverse research practices. Additionally, we reconsidered the five-star model of Open Data in humanities research, where full open licensing or RDFbased linking is often impractical. We propose a revised model separating the provenance, licensing, and external linkage from core textual structuring. Our principle aligns with Levels 2–4 of the Best Practices for TEI in Libraries, but with lower overhead due to focused, purpose-driven markup. By grounding structuring in research needs, the Tidy Text approach aims to support incremental, but interoperable documentation pathways, thereby broadening the reach of TEI in the digital transformation of scholarly practice. Streamlining TEI Workflows Collaborative Editing with NormaTEI Pietro Sichera (1); Salvatore Cristofaro (1); Christian D’Agata (2); Angelo Mario Del Grosso (1); Miryam Grasso (2); Laura Mazzagufo (1); Daria Spampinato (1) (1) Consiglio Nazionale Delle Ricerche – CNR, Italy; (2) Università degli Studi diCatania, Italy Keywords Cooperative Digital Scholarly Edition, Harmonization, Encoding, Interoperability, NormaTEI Abstract In digital philology, the production of scholarly editions is increasingly recognized as a collaborative endeavour, involving multidisciplinary teams and both synchronous and asynchronous workflows. This shift affects not only the publication of individual critical texts but also the construction of large-scale digital archives. While adopting formal protocols like those of the TEI is essential, it alone cannot guarantee consistency and coherence of the outcomes. A robust system is needed to support the entire editorial process, facilitating both the harmonization and quantitative analysis of texts. LONG PAPER 190 The CMIFerator. A generalised pipeline… The CMIFerator. A generalised pipeline… 191 The CMIFerator A generalised pipeline for contributing to correspSearch Julian Jarosch Academy of Sciences and Literature Mainz, Germany Keywords CMIF, correspSearch Abstract During the work on long-running digital scholarly editions of letters, it may become desirable to periodically generate and update Correspondence Metadata Interchange Format (CMIF) (TEI Correspondence SIG, 2025) files for use in correspSearch (Dumont et al., 2025). This may typically be based on a non-finalised working set of data, on data conforming to a superset of the CMIF schema (containing extraneous attributes or elements not covered by CMIF version 1.0), and/or on normalised data distributed over many individual letter files plus indices of persons, places and organisations (i.e. the structure required by ediarum (TELOTA, 2025). From the observation of these requirements in a number of projects (Daugirdas and Kuczera, 2025; Dingel and Kohnle, 2023; Trautmann and Schrade, 2018; Wiese, 2025), the CMIFerator (Jarosch, 2025), a function library for eXist-db, was developed. It provides a set of functions for converting TEI letter files to CMIF XML. The design as a function library provides SHORT PAPER some flexibility regarding which parts of the transformation are needed in the individual context. The full pipeline offered by the CMIFerator is: • Update <correspAction> elements in individual letter files with the most up-to-date information from index files (regularised person names, person identifiers, regularised place names …). • Subset <correspDesc> elements in individual letter files for (strict) conformance to the CMIF standard. • Wrap <correspDesc> elements in a CMIF template and fill in metadata. The full pipeline is made accessible through convenient wrapper functions. At the same time, the three individual steps are available as separate functions and can be used individually or selectively if necessary. The project-specific configuration is covered by a lean configuration file. At present, the CMIFerator intentionally is not conceived as a plug-and-play app, but instead example configurations, transformations and scripts are provided. These examples demonstrate how to use the function library in an XQuery script to either generate CMIF on the fly, or store it in the eXist database. References Daugirdas, K., Kuczera, A. (Eds.), 2025. Zwischen Theologie, frühmoderner Naturwissenschaft und politischer Korrespondenz: Die sozinianischen Briefwechsel. Akademie der Wissenschaften und der Literatur, Mainz. Dingel, I., Kohnle, A. (Eds.), 2023. Flacius Briefwechsel. Digital. Akademie der Wissenschaften und der Literatur, Mainz. The TEI Critical Apparatus at 30-something… 193 Dumont, S., Grabsch, S., Müller-Laackman, J., Sander, R., Sobkowski, S., 2025. correspSearch – Briefeditionen vernetzen. Jarosch, J., 2025. CMIFerator. TEI Correspondence SIG, 2025. Correspondence Metadata Interchange Format (CMIF). TELOTA, 2025. ediarum – Digitale Editionen erstellen und publizieren. Trautmann, M., Schrade, T. (Eds.), 2018. DER STURM. Digitale Quellenedition zur Geschichte der internationalen Avantgarde. Akademie der Wissenschaften und der Literatur, Mainz. Wiese, C. (Ed.), 2025. Buber-Korrespondenzen Digital: Das Dialogische Prinzip in Martin Bubers Gelehrtenund Intellektuellennetzwerken im 20. Jahrhundert. Akademie der Wissenschaften und der Literatur, Mainz. The TEI Critical Apparatus at 30-something Where do we go from here? Hugh Cayless Duke University, USA Keywords Critical Apparatus, Digital Critical Editions, TEI Guidelines Abstract The TEI’s Critical Apparatus chapter seems to have first appeared in the TEI P3 Draft, published in print in 1994 (it is mentioned earlier in the P2 sources in the TEI Vault, but is not present there). It had then most of the features it has now, the app, lem, rdg, rdgGrp, wit, witDetail, the associated witStart/ End and lacunaStart/End as well as witList (now listWit) and witness. Much of the prose and many of the examples in the chapter date back to that period. Substantial changes were made in 2015 and 2020 to permit the modeling of structural variation, tighten up the content model, and to improve the discussion of critical practices that the apparatus module may be used to implement. Despite these improvements, however, the Critical Apparatus chapter remains somewhat confusing and disjointed. One example where the chapter takes a turn into obscurity is the section on “subvariation” (13.1.3) which introduces the LONG PAPER 194 The TEI Critical Apparatus at 30-something… Towards a model of transcultural encoding of ancient… 195 rdgGrp element and explains how it might be used to subdivide readings for the beginning of Chaucer’s Wife of Bath’s Tale. It should be noted that printed critical editions of Chaucer make little of this example, printing ‘Experience’ and noting the variant ‘Experiment’. The reader is left wondering why they might commit to such elaborate encoding. A far better and more understandable use case for rdgGrp would be the encoding of variorum editions, where it can quite logically be used to collect the apparatus entries from previous editions of the work. Part of the reason for the chapter’s state is simply its age, and a reluctance on the part of the Guidelines’ editors to commit to a thorough reworking, along with a certain deference to its original creators. Part, too, is the difficulty of the material. Further, the apparatus module is complicated by its attempt to do many things at once. It can model variation on a text, it can serve as a collation table, it can act as connective tissue between a set of varying sources, or it can serve to transcribe a previously printed apparatus. It features no less than three techniques for attaching the apparatus to the text, two of which are arguably obsolete in the P5 era, and none of which account adequately for the situation where an editor is presenting one apparatus with many texts. Moreover, as noted above, some of the examples recommend techniques for which it is hard to understand the desired outcome, being more theoretical than practical. For all of these reasons, it is time for a re-consideration of the critical apparatus chapter, to prune some of the outdated recommendations and examples and to better account for recent work on digital critical editions. My talk will propose a new framework for organizing the chapter, centered around four basic use cases: single text critical editions, multitext critical editions, collation tables, and the encoding of existing print editions. It will suggest the chapter consider only two forms of apparatus connection: inline (i.e. parallel segmentation) and external. Towards a model of transcultural encoding of ancient epigraphic sources Michele Brunet (1); Estelle Ingrand-Varenne (2); Nicolas Souchon (5); Nathalie Prévôt (3); Vincent Razanajao (6); Coline Ruiz-Darasse (3); Blandine Nouvel (4); Emmanuelle Morlock (1); Bruno Baudoin (4); Satre Stéphanie (4); Rémi Bonnin (2); Léontine Fortin (2); Damien Strzelecki (2) (1) Université Lumière Lyon 2, France, CNRS UMR 5189 HiSoMA, EquipEx Biblissima+; (2) CNRS UMR 7302 CESCM, EquipEx Biblissima+; (3) CNRS UMR 5607 Ausonius, Université Bordeaux Montaigne; (4) CNRS, UMR 7299 Centre Camille Jullian, Aix Marseille Université; (5) Institut français d’archéologie orientale; (6) Centre d’études alexandrines, CNRS-IFAO Keywords Epigraphy, TEI, EpiDoc, Opentheso, Patrimonium Editor. Abstract This paper presents the work of the 5A cluster, “TEI and Epigraphy”, within the Biblissima+ consortium (Observatory of Written Cultures, from Clay to Print). In order to better capture the semantic interrelations between artefacts, images, and texts in epigraphic sources from Ancient Egypt, Greece, Rome, Gaul, and the Middle Ages, we have developed an integrated LONG PAPER 196 Towards a model of transcultural encoding of ancient… digital environment. This environment combines a generic XML/ TEI/EpiDoc template (GenEpiTemplate) with a structured multilingual thesaurus (GenEpiTheso), both accessible through a dedicated web-based editorial platform. Biblissima+ provides a digital library and research infrastructure designed to support the study of the transmission of texts from Antiquity to the Modern era. As part of its broader scholarly objectives, the consortium initiated the integration of epigraphic data. An interdisciplinary team — comprising epigraphers working across diverse cultural and chronological contexts, together with specialists in Digital Humanities — was assembled to investigate unified methods of data representation. This cross-cultural approach contributes to refining the conceptual definition of the epigraphic document, emphasising its material, graphic, and iconographic dimensions. The initial aim of the cluster was to establish a shared editorial template for its various projects, all of which rely upon the standards of the Text Encoding Initiative (TEI) and EpiDoc. Although a core model was rapidly identified, the first version revealed a number of limitations. Through comparative analysis of existing encoding practices, we developed an enriched and fully annotated template, extending EpiDoc with a broader range of TEI elements to accommodate a more extensive corpus of epigraphic material. Concurrently, we developed a multilingual thesaurus, organised into seven thematic domains (artefact, material, text, language, script, production, and field of study). Each entry is accompanied by definitions, translations, and conceptual alignments, and is managed using the open-source platform Opentheso. Both the encoding template and thesaurus are now accessible via the PETRAE portal and have been integrated into the PATRIMONIVM editor — an eXist-db application developed by the Ausonius Institute — facilitating the online editing of XML/ TEI-EpiDoc encoded inscriptions. The platform is further connected to a Zotero library (GenEpiBiblio), which consolidates essential bibliographic references across the principal domains of epigraphic research. This presentation will introduce the project and its principal outcomes, and wishes to open a discussion with TEI specialists concerning the methodological and technical challenges encountered during its development. 198 Towards an Encoding Practice for Multilingual Textual Variation Towards an Encoding Practice for Multilingual Textual Variation 199 Towards an Encoding Practice for Multilingual Textual Variation Sandra Balck; Fabian Etling Freie Universität Berlin, Germany Keywords digital scholarly edition, multilingualism, text comparison, textual criticism Abstract Multilingual text versions pose challenges for digital scholarly editions (DSE), particularly in the encoding and representation of textual variation. Hannah Arendt for example authored most of her works in English and German, carefully rewriting and editing the texts herself. The resulting versions should not be considered as mere translations or revisions. Arendt used the plurality of languages as a creative tool. In comparing the versions, seeming textual shifts create meaningful tensions. For a DSE, this imposes the task of enabling readers of Arendt’s works to explore “these tensions as a significant and signifying Between.” (Wild, 2024, p. 29) The COMUTE project 9 was initiated to analyse such “tensions” in multilingual writing. Funded by the German Research Foundation, it aims to algorithmically identify variant passages 10 9 https://www.comute-project.de/ 10 The COMUTE-Project is building on the existing monolingual collation tool LERA (Pöckelmann et al., 2022) and is extending it by adding a multilingual layer. POSTER and, in collaboration with the Hannah Arendt Critical Edition11 as one use case, explores how such findings can be evaluated, visualised and integrated into editorial workflows. In order to support data reuse, especially in DSE workflows, the collation results should be encoded according to the TEI guidelines.12 While it is feasible to use TEI markup for critical apparatuses for simpler cases, this is not straightforward when it comes to more complex cases, as phenomena of multilingual textual variation manifest themselves not only at the character level, but also in linguistic nuances and in content and context. This poster13 presents a TEI-based approach for encoding multilingual and non-parallel textual variation and integrating it into DSEs. References Bleeker, E., Buitendijk, B., Haentjens, R. and Kulsdom, A. (2018) ‘Including XML Markup in the Automated Collation of Literary Text’, XML Prague 2018 Conference Proceedings, pp. 77–95. Available at: https://archive.xmlprague.cz/2018/files/xmlprague-2018-proceedings.pdf Cayless, H., Beshero-Bondar, E., Vigilanti, R. and Cummings, J. (2019) ‘Document Modeling with the TEI Critical Apparatus’, TEI 2019: What is text, really? TEI and beyond, University of Graz, Austria, 16–20 September. Book of Abstracts, pp. 168–170. Available at: https://gams.uni-graz.at/o:tei2019. bookofabstracts 11 https://hannah-arendt-edition.net/home 12 As discussed for monolingual variation, for example, by Bleeker et al. (2018) and Cayless et al. (2019) and for translation scenarios in the TEI mailing list (Various Authors, 2013). 13 DOI: https://doi.org/10.5281/zenodo.15834844 Voci dall’Inferno. a TEI-Based Digital Archive… 201 Pöckelmann, M., Medek, A., Ritter, J. and Molitor, P. (2023) ‘LERA – An interactive platform for synoptical representations of multiple text witnesses’, Digital Scholarship in the Humanities, 38(1), pp. 330–346. Available at: https://doi.org/10.1093/llc/ fqac021 Various authors (2013) ‘Translation as witness’, TEI (Text Encoding Initiative) public discussion list, 6 November 2013 13:56:01 - 0500. Archive of thread available at: https://lists. psu.edu/cgi-bin/wa?A2=1311&L=TEI-L&D=0&P=15199371 Wild, T. (2024) ‘Relational Reading: Hannah Arendt’s The Human Condition and Vita activa, and The Plurality of Languages’, Angermion, 17(1), pp. 139–158. Available at: https://doi.org/10.1515/ anger-2024-0005 Voci dall’Inferno a TEI-Based Digital Archive for finding Dante in Concentration Camp Testimonies Elvira Mercatanti (1); Marina Riccucci (2); Angelo Mario Del Grosso (1) (1) ILC: CNR-Istituto di Linguistica Computazionale “A. Zampolli”, Pisa, Italy; (2) Università di Pisa, Italy Keywords XML-TEI, digital archives, eXist-db, XQuery, web application Abstract Voci dall’Inferno is a digital humanities project that brings together a multidisciplinary team of scholars and Digital Humanities students. The project pursues two interconnected objectives: (a) the creation of the first digital corpus of non-literary testimonies of concentration camp survivors – mostly unpublished – encoded in XML-TEI, including both written and oral sources (Burnard, 2014); (b) the systematic identification and analysis, within the testimonies, of lexical items and allusions to Dante’s Divine Comedy, particularly the Inferno (Calderini and Riccucci, 2020; Mattingly, n.d.). The project is grounded in the hypothesis that Dante’s lexicon provided survivors with an expressive framework to talk about the ineffable experience of the Lager, overcoming the limits of its own personal vocabulary. The frequent and meaningful recurrence of Dantean references – regardless of educational POSTER 202 Voci dall’Inferno. a TEI-Based Digital Archive… Voci dall’Inferno. a TEI-Based Digital Archive… 203 background – suggests the Commedia acts as a shared cultural code (Calderini and Riccucci, 2020). The ongoing digital corpus is accessible through a web application developed within the eXist-db environment using XQuery, which supports querying, annotation visualization and intertextual analysis (see a demo at: https://tinyurl.com/ mkcez3f8). Encoding strictly follows profitably TEI Guidelines, employing modules for encoding oral data and modules for primary written sources as well as named entities and an extensive metadata recording. A parallel transcription approach was adopted: TEI zone elements for written sources and TEI timeline mechanisms for oral ones. The corpus is developed in line with FAIR principles (Del Grosso et al., 2024b). The archive includes 23 testimonies from 18 witnesses (13 oral, 10 written), totalling 18 hours and 35 minutes of audio and 390 transcribed pages. Preliminary analysis reveals 61 references to the Comedy, highlighting how Dante’s Inferno serves as a semantic bridge between literary memory and the survivors’ needs (Fig. 1) (Del Grosso et al., 2024a). Figure 1: distribution of Dante’s lexicon across the corpus: a total of 61 references, including 15 explicit quotations, 4 implicit quotations, 7 allusions, and 35 terms. Our current work focuses on the automatic text recognition task, using tools like Whisper and eScriptorium, and the automatic extraction of Dante’s quotations (Fig. 2), using SentenceBERT and VectorializedDB. Figure 2: Voci dall’Inferno Verse Similarity Search: a sentence from a testimony is used as input, and the output returns the verses from Dante’s Divine Comedy that are most semantically similar. This paper will present the TEI encoding strategies adopted as well as the broader methodological implications of the project. 204 Voci dall’Inferno. a TEI-Based Digital Archive… What do we Need to Make Documents from Texts? 205 References Burnard, L. (2014). What is the Text Encoding Initiative? How to add intelligent markup to digital resources. Marseille: OpenEdition Press. https://doi.org/10.4000/books.oep.426 Calamai, S., Beeken, J., Scagliola, S., Broekhuizen, M., Draxler, C., Van Hessen, A. and Van den Heuvel, H. (2021). ‘Voices from Ravensbrück: Towards the creation of an oral and multilingual resource family’. In: Proceedings of CLARIN Annual Conference 2021, pp.16–19. Calderini, S. and Riccucci, M. (2020). ‘L’ineffabilità della nefandezza: Dante “per dire” il Lager: un sondaggio preliminare nelle testimonianze non letterarie’. Italianistica, 49, pp.213–228. https://doi.org/10.19272/202001301011 Del Grosso, A.M., Riccucci, M. and Mercatanti, E. (2024a). ‘Voci dall’Inferno: Dante per dire il Lager - Digitalizzare e studiare le testimonianze’. In: ME.TE. Digitali: Mediterraneo in rete tra testi e contesti. Catania: AIUCD. Del Grosso, A.M., Riccucci, M. and Mercatanti, E. (2024b). ‘The impact of digital editing on the study of Holocaust survivors’ testimonies in the context of Voci dall’Inferno project’. In: Anuradha, I., Wynne, M., Frontini, F. and Plum, A. (eds.) Proceedings of the First Workshop on Holocaust Testimonies as Language Resources (HTRes) @ LREC-COLING 2024. Torino, Italia: ELRA and ICCL, pp.1–9. https://aclanthology.org/2024. htres-1.1/ Levi, P. (1997). Conversazioni e interviste (1963–1987). Belpoliti, M. (ed.). Torino: Einaudi. Mattingly, W. (n.d.). wjbmattingly/vulgata-spacy. [Online] Available at: https://github.com/wjbmattingly/vulgata-spacy (Accessed 26 June 2025). Streamlit (n.d.). Latin Vulgate App. [Online] Available at: https:// latin-vulgate.streamlit.app/ (Accessed 26 June 2025). What do we Need to Make Documents from Texts? Georg Vogeler University of Graz, Austria Keywords documentality, authentication, diplomatics Abstract The term document appears frequently in the TEI Guidelines, generally with three meanings: (1) any textual object (e.g., in <docTitle>), (2) a computer file containing user information (e.g., the <TEI> element), and (3) a material object bearing conventional symbols (e.g., in <sourceDoc>). This usage reflects general language, as captured by dictionaries like Merriam-Webster. Within the TEI community, and especially among digital scholarly editors, a narrower focus has prevailed, centered on the third definition and practices associated with “documentary editing.” A broader perspective is offered by Maurizio Ferraris, who proposes a theory of documentality—seeing documents as fundamental social artefacts that define social reality (Ferraris 2009, 2011, Ferraris & Torrengo 2014). This view aligns with the historical discipline of diplomatics, which studies methods developed to authenticate and manage social relationships through documents. Diplomatics has thus become both a field of historical forensics (discrimen veri ac falsi) and a cultural history of documents. Luciana Duranti (1989, 1998) has continued this tradition into modern archival science. LONG PAPER 206 What do we Need to Make Documents from Texts? What do we Need to Make Documents from Texts? 207 Central to diplomatics is the authenticity of documents: establishing verifiable links between a text and the individuals creating the social relationship it records. Historically, this was achieved through public acts, reliance on trusted officials like notaries, the development of stylistic norms (e.g., in papal chancery), and material signs like signatures and seals. Today, digital methods such as hashes and cryptographic signatures fulfil similar roles. Sean Winslow (2020) has proposed how TEI encoding could better capture these features. Meanwhile, the Vocabulaire international de la diplomatique (VID, Cárcel Ortí 1997)14 provides an established terminology for describing these aspects, showing their relevance across cultures (Vogeler 2018). However, the TEI Guidelines currently lack an ontological model for representing the documentary status of texts in the diplomatic sense. Although diplomatics documents have been encoded with TEI, often semantically reduced (e.g., Isasi Martínez & Spence 2014; MRSH Caen15), or by integrating the VID (as in Diple16 or Vujosevic & Vogeler 2025), these efforts remain marginal. The Charters Encoding Initiative (CEI)17 represents a more formal attempt but is based on TEI P4. Yet these initiatives expose a deeper issue: the TEI’s conceptual model does not recognize the full social and forensic nature of documents. Diplomatic concepts like the authenticum or copy chains crucial to establishing authenticity are poorly represented. TEI’s <signed> and <seal> elements encode textual and physical features without connecting them to their authenticating functions. <filiation> remains purely relational. In 2022, Sean Winslow submitted a formal proposal to better accommodate diplomatic concepts in TEI.18 While the commu14 https://www.cei.lmu.de/VID/ 15 https://mrsh.unicaen.fr/editer-des-sources-diplomatiques/ 16 http://developpements.enc.sorbonne.fr/diple/schema/acte/ 17 https://www.cei.lmu.de 18 https://github.com/TEIC/TEI/issues/2376 nity’s cautious response—suggesting, for instance, the use of customization profiles—is understandable, I argue that this reluctance reflects a deeper neglect of essential textual concepts. In my presentation, I will advocate for rather small extensions to the teiHeader to represent the documentary and forensic dimensions of texts: a new <authDesc> as property of the TEI document or an individual textual witness. <authDesc> can include descriptors of means of authentication like seals, stamps, signatures, md5 hashes and similar or a prose reflection on the authenticity status. A controlled vocabulary for the core authenticity status (original, forged, suspicious, unclear) supports machine readability. An element focussing on the roles in text creation and processing that support its documentary status as well as the role in the social act (<legalActor>) support this and can be detailed by referencing established external sources like mentioned Vocabulaire Internationale de la Diplomatique. I invite the TEI community to discuss whether we should simply accept the status quo—or whether we can, and should, open the TEI Guidelines to communities focused on documentality: digital archivists, legal scholars, and—certainly—students of historical documentary cultures. References Cárcel Ortí, Maria Milagros. 1997. Vocabulaire International de La Diplomatique. 2nd ed. Collecció Oberta. Valencia: Univ. de València. Duranti, Luciana. 1989. ‘Diplomatics : New Uses for an Old Science’. In Archivaria 28, 28:7–27. Ottawa. https://www.academia. edu/5404671/Diplomatics_New_Uses_for_an_Old_Science. –. 1998. Diplomatics: New Uses for an Old Science. Lanham: Scarecrow Press. Ferraris, Maurizio. 2009. Documentalità. Perché è Necessario Lasciar Tracce (Eng. Tr. by R. Davies, Documentality: Why It Is 208 What do we Need to Make Documents from Texts? Who Knows What a Revision Is? 209 Necessary to Leave Traces, New York, Fordham University Press, 2012). Roma-Bari: Laterza. —. 2011. ‘Social Ontology and Documentality’. In Approaches to Legal Ontologies, edited by Giovanni Sartor, Pompeu Casanovas, Mariangela Biasiotti, and Meritxell FernándezBarrera, 83–97. Law, Governance and Technology Series 1. Springer Netherlands. http://link.springer.com/ chapter/10.1007/978-94-007-0120-5_5 Ferraris, Maurizio, and Giuliano Torrengo. 2014. ‘Documentality: A Theory of Social Reality’. Rivista Di Estetica 57: 1 1 – 27. https:// doi.org/10.4000/estetica.629 Isasi Martínez, Carmen, and Paul Spence, eds. 2014. ‘Guía Para Editar Textos CHARTA Según El Estándar TEI: Una Propuesta.’ https://www.redcharta.es/investigacion/ Vogeler, Georg. 2013. ‘Von der Terminologie zur Ontologie : Das “Vocabulaire international de la diplomatique” als Ressource des Semantic Web’. Francia 40:281–97. Vogeler, Georg. 2018. ‘Digital Diplomatics: The Evolution of a European Tradition or a Generic Concept?’ In Studies in Historical Documents from Nepal and India, edited by Simon Cubelic, Astrid Zotter, and Axel Michaels, 1:85–109. Documenta Nepalica. Heidelberg: Heidelberg University Publishing. https:// doi.org/10.17885/heiup.331.c4129 Vujošević, Žarko, and Georg Vogeler. 2025. ‘Encoding Medieval Charters in Preparation for a Diplomatic Semantic Web. Combining the TEI with the Vocabulaire International de Diplomatique’. Zenodo. https://doi.org/10.5281/ zenodo.14696237 Winslow, Sean M. 2020. ‘Authenticating Features in the TEI’. Journal of the Text Encoding Initiative, no. Issue 13 (May). https://doi.org/10.4000/jtei.3608 Who Knows What a Revision Is? Towards a Shared Vocabulary of Textual Variation Elli Bleeker (1); Beatrice Nava (2) (1) The Huygens Institute, Netherlands; (2) University of Vienna, Austria Keywords textual variation, collation, visualization, ontology, textual scholarship Abstract This paper presents preliminary results from an initiative introduced at the international workshop Seeing the Difference: Visualizing Textual Variation (Leiden, April 2024), aimed at aligning scholarly vocabularies used to describe textual variation. While the idea of a mapping of textual scholarship terminology is not new (Dillen 2020; Macé and Roelli, eds.), the need remains. For example, the understanding of “revision” differs across historical periods and material contexts, despite using the same TEI-encoding. The great advantage of the TEI is that it offers textual scholars flexibility in recording and categorizing variation, however, this lack of standardization is also a core problem for the development of tools and visualisations of textual change, as it prevents interoperability, replicability and evaluation of results across textual traditions. Therefore, consensus on vocabulary and specifications for functional SHORT PAPER