Introduction

While East Asian digital humanities has witnessed substantial infrastructure development in Japan through the Aozora Bunko (Aozora Bunko, 1997–) and in China via the Chinese Text Project (Sturgeon, 2021), Korean literature remains conspicuously absent from the global TEI ecosystem. This gap is not merely quantitative but infrastructural and methodological: Korean texts present distinct challenges including agglutinative morphology, historical orthographic variations (old Hangeul), mixed-script conventions with Chinese characters (hanja), and colonial-period narrative strategies that existing TEI frameworks inadequately address.

This half-day workshop presents a complete, replicable pipeline for building a TEI/XML database of Korean literature, demonstrated through a corpus of 30 works by 16 canonical authors spanning from the 1910s to 1940s. The corpus encompasses the full arc of Korean modern literary history—from Lee In-jik's "new novel" (sinsoseol) through the naturalism of Hyun Jin-geon, the socialist realism of the KAPF movement, to the high modernism of Yi Sang. Notably, it includes works by five female authors (Na Hye-seok, Kim Myeong-sun, Baek Sin-ae, Kang Kyeong-ae, Chi Ha-ryeon) and diasporic literature by the Zainichi Korean writer Kim Sa-ryang. By the end of this workshop, participants will be able to independently harvest, encode, and analyze Korean literary texts using the complete pipeline presented here.

Session 1: Harvesting Korean Literary Texts

This session introduces a hybrid approach combining API-based harvesting with dump file supplementation. We demonstrate how to scrape Korean Wikisource, extract metadata via the Wikimedia API, and enrich records by cross-referencing the Encyclopedia of Korean Culture API (Academy of Korean Studies, 2024). The resulting workflow produces TEI-ready data with <sourceDesc> structures linking Wikisource IDs, Wikidata Q-numbers (e.g., Q61097625), and encyclopedia references—ensuring provenance transparency and semantic interoperability.

Session 2: Building a TEI Auto-Tagging Application with LLM

This session demonstrates how to build a web-based TEI auto-tagging application using Gemini in Google AI Studio. Rather than merely presenting results, we walk participants through the entire app development process: designing effective system prompts that enforce TEI P5 compliance (TEI Consortium, 2025), handling Korean-specific challenges such as agglutinative morphology and hanja transliteration, and integrating external knowledge bases for metadata enrichment. The fully functional application will be publicly released, enabling participants to immediately apply automated TEI conversion to their own Korean text collections.

Session 3: Tools for TEI Markup

Even when Large Language Models generate initial TEI drafts, careful human review remains essential. This session surveys TEI editing tools, including web-based editors and VS Code extensions, with attention to Rahtz Prize-winning applications. We compare tool strengths for different workflow stages—structural tagging, attribute assignment, validation—and provide hands-on practice applying markup to sample texts drawn from Yadam. Participants will learn practical strategies for efficient error detection and batch correction across large text collections.

Session 4: A Case Study on Encoding Old Hangeul Novel Texts

Using Lee Hae-jo's Hwauihyeol (1911) as a case study, this session explores the encoding lifecycle for early twentieth-century Korean novels. Hwauihyeol, written in Old Hangeul and serialized in the Korea Daily News (“Hwauihyeol”, 2024), exemplifies one of the ways in which early twentieth-century Korean novels were historically published and can be digitally encoded. We begin with the identification of original scripts in digitized texts and then introduce the multiple ways of annotating entities relevant to the theme of the text. Finally, we demonstrate the importance and feasibility of encoding Old Hangeul novels with the annotated results from other early twentieth-century Korean novels, such as Lee In-jik’s Hyeoruinu (1906) and Lee Hae-jo's Cheolsegye (1908). Building on prior work (Burnard, 2001; Ham, 2018; Lee, 2024), we argue, following Burnard (2001), that TEI encoding is an "act of interpretation."

Session 5: Collaborative Encoding and a Korean Literary Model

This session presents encoding guidelines developed through collaboration among eleven contributors. A central challenge is representing diverse narrative and stylistic features in Korean literature of the 1910s-1940s; through case studies including Na Hye-seok's Kyeonghui (1918), Kim Nam-cheon's Mul (1933), and Ji Ha-ryeon's Dojeong (1946), we demonstrate how our system addresses how our system addresses (1) interior monologue and speech acts; (2) orthographic instability and code-switching between Korean, Japanese, and hanja (Sino-Korean characters); and (3) period-specific ideological terminology and historical contextualization. We then compare our approach with Japanese literary TEI projects, particularly Aozora Bunko's structural conventions, proposing a model that addresses Korean literature's unique features while maintaining East Asian interoperability.

Session 6: From TEI Data to AI-Enhanced Reading

Why invest in high-quality TEI encoding? This concluding session demonstrates how structured literary data enables new research methodologies through Retrieval-Augmented Generation (RAG). Drawing on the Japanese Humanitext project (Ogawa et al., 2025), we illustrate how well-encoded corpora support AI-driven literary analysis: character network extraction, speech pattern analysis, and cross-textual thematic exploration. The session articulates a vision wherein TEI encoding yields dividends for both traditional scholarship and emergent computational approaches.

Conclusion

This workshop represents the first systematic effort to establish TEI-based infrastructure for Korean literature. Our corpus of 30 works—featuring five female authors and encoded collaboratively by eleven contributors including senior researchers and graduate students—demonstrates that the proposed pipeline is both scalable and replicable. By sharing our encoding guidelines, tool configurations, and sample files through a publicly accessible GitHub repository, we aim to catalyze an East Asian literary TEI community and offer a model for non-Western literary traditions entering the global digital humanities infrastructure.

Workshop Information

Format: Half-day workshop (3 hours), hybrid in-person and online participation supported

Target Audience: Digital humanities researchers, Korean literature scholars, East Asian studies specialists, librarians and archivists, graduate students. Basic XML familiarity helpful but not required; no prior TEI experience necessary.

Expected Participants: 15–25

Language: Korean/English presentations with Korean text examples (romanization provided)

Technical Requirements: Participants should bring laptops with VS Code installed. Workshop materials, sample files, and tool configurations will be distributed via GitHub repository prior to the session.

Diversity Statement: This workshop foregrounds a non-Western literary tradition underrepresented in global DH infrastructure. Our corpus includes five female authors and represents ideological diversity spanning nationalist, socialist, modernist, and feminist perspectives, as well as diasporic literature. The workshop team includes both senior researchers and graduate students with gender-balanced representation.