Projects

A large corpus of New High German poetry, built by collecting and parsing the bulk of digitized public-domain German verse — chiefly the Deutsches Textarchiv and the TextGrid Digital Library. The current v6 holds ~65,700 poems by ~250 authors (over 12 million tokens), segmented to syllable level. It underpins much of my distant-reading work on verse.

A corpus of German and English poetry annotated line-by-line for the aesthetic emotions it evokes (beauty/joy, sadness, suspense, awe, and more), with prosody annotation and modeling code — 158 German and 64 English poems double-annotated, plus a Chinese pilot. The dataset behind the PO-EMO paper.

German and English poetry corpora annotated with prosodic features — syllable stress, meter, measure, caesura — together with the neural taggers trained on them. Code and data for the EACL 2021 paper.

A diachronically balanced sample of German poetry, manually annotated for rhyme — notably, about a third of stanzas don’t rhyme at all. Encoded in TEI P5; the resource behind the supervised rhyme-detection paper (SIGHUM 2018).

Tools for scraping, cleaning, and assembling large poetry corpora from heterogeneous sources.

An English poetry corpus mined from Project Gutenberg (via GutenTag) and prepared for computational analysis.

A diachronically balanced, hand-selected gold corpus of English poetry for evaluating automatic analysis.

Canonical German poems paired with student interpretations and several annotation layers, crawled from the Antikörperchen poetry site. (See also the related “antik” repository.)

A small reference set of English poems with manual gold annotation.

Parsers for reading poetry encoded in TEI/DTA-conforming XML into analysis-ready formats. (See also “poetry-api-xml-dta”.)

Data and code behind “The Ongoing Birth of the Narrator” (DSH 2025): annotation guidelines, the gold-labeled instances, and the model folds used to trace the emergence of the author–narrator distinction in literary criticism.

A dataset of titles and authors on the New York Times fiction bestseller list, 2000–2020 — the basis for the “More Social, Less Religious” trend study.

A corpus of over 110,000 erotic fan-fiction documents crawled from literotica.com, organized by user rating and genre — assembled for computational study of genre, style, and narrative.

A set of web crawlers used to build text corpora — for the Antikörperchen poetry site, Literotica, and MGG-Online (musicology).

Scripts that turn TEI-encoded XML into the printed PDF Book of Abstracts for the DHd2024 conference in Passau, which I co-organized (adapted from a TEI-to-PDF pipeline).

A corpus and processing pipeline for German parliamentary proceedings (Bundestag Plenarprotokolle), underpinning our work on linking parliamentary speakers to Wikidata and on German nuclear-energy discourse. (A collaboration with M. Teich; repository internal.)

Citation, co-authorship and keyword network analysis of the DHd conference abstracts (2014–2023) — the code behind “Quo Vadis Fachbereiche und Schulen der DHd.” (A collaboration with S. Gassner; repository internal.)

A curated collection of psycholinguistic valence (affective pleasantness) rating datasets — 37 datasets across 13 languages, mapped to Concepticon via NoRaRe — used to evaluate LLM valence scaling with Best-Worst Scaling across 16 models (0.5B–671B parameters). The basis of the forthcoming valence-norms paper.

A pipeline that extracts plain text from PDFs and uses a local Mistral model to pull book metadata — and, soon, literary mentions — from a corpus of secondary literature on ecocriticism, toward mapping the “ecocritical canon.”

Code and data for retrieving entities and texts from the EcoCor API — part of the open EcoCor infrastructure for digital ecocriticism introduced at DH2026.

Code and data for detecting enjambement — line breaks that run against syntax — in German and Spanish poetry. Supports the forthcoming CHR paper. (Repository is internal.)

Modeling and prediction over the EcoCor ecocriticism corpus. (Repository is internal.)

Modeling experiments around ecocriticism and the environmental humanities. (Repository is internal.)

Earlier experiments on automatically scaling affective (valence) norms for words. (Repository is internal.)

(Description to come.)