Daejeon, July 27–31
Over the past decade, environmentally conscious scholarship has taken on a growing presence in Digital Humanities (“DH”) (Posthumus, Sinclair 2014). Under labels such as “Digital Environmental Humanities” (Jørgensen 2014, Posthumus et al. 2018, Langer et al. 2022, Ryan et al. 2023) or “Ecocritical DH” (Ladino 2018), researchers explore how DH can help addressing the ecological multi-crisis of our time through digital methods, complementing the research in Environmental Humanities. In this vein, DH seeks to contribute to what might be termed ‘Ecological Enlightenment‘: an anthropocritical inquiry into the history of human-environment relations, the activities of ‘more-than-human’ agents, and the epistemological and moral consequences of such inquiry for our present. When it comes to the digital study of literature, what we call “Digital Ecocriticism” builds on theoretical positions within Environmental Humanities that reframe cultural texts not as human-centred documents but as assemblages of beings, materials, and relations. Latour’s (2005) redefinition of collectives as hybrid constellations of humans and non-humans that co-produce social worlds and Haraway’s (2016) approach to understanding the world through multispecies entanglements resonates strongly with this kind of scholarship that seeks to make visible the ‘more-than-human’ agencies that shape not only history but also our stories and “naturecultures” (Haraway 2003).
From a systematic perspective, three interrelated areas of action can be identified for Digital Ecocriticism:
In this contribution, we introduce EcoCor, an open infrastructure for the presentation, exploration, and analysis of corpora with ecocritically relevant literary texts. EcoCor is currently both a conceptual framework, currently discussed within a growing, interdisciplinary collective of scholars, and an operating prototype of an actual DH infrastructure, first developed during the Eco-Hackathon "Environmental Data, Media and the Humanities" (Potsdam, 2023) and later enhanced at the follow-up Eco-Hackathon "Environments in and as Networks" (Potsdam, 2026).
Crucial to the concept of EcoCor is the idea of reusing existing infrastructural components to ensure a sustainable approach to DH tech development. In line with this idea, EcoCor's tech stack is based on that of the DraCor platform. Along with the software architecture, we adopt for EcoCor the concept of programmable corpora, which is the guiding principle behind the development of DraCor (Fischer et al. 2019; cf. Börner & Trilcke 2023, Börner & Trilcke 2025). The most important components of EcoCor are accordingly: 1) the corpus base, 2) microservices, 3) a frontend and 4) an API, the latter of which opens up the corpora for machine interaction and thus makes them ‘programmable’. We briefly report on each of these components.
In the current prototype phase, EcoCor hosts an English and a German corpus. The English corpus currently comprises a combined collection of American English and British English texts. Corpora in other languages will be contributed by other corpus maintainers in a community-driven approach following the principle of collaborative corpus building.
Technically, EcoCor consists of curated collections of XML/TEI documents. For the encoding of the TEI files, we use the guidelines and schemata of the ELTeC project (Schöch et al. 2021). Cf. https://www.distant-reading.net/ and https://github.com/COST-ELTeC
At the 2026 Eco-Hackathon, we successfully completed the EcoCor100 challenge for the German-language EcoCor corpus and developed and published an initial release version of the corpus containing a total of 183 texts (Schuhmacher et al. 2026). We plan to expand the English-language corpus beyond the 100-text threshold as well. Additionally, an Italian and a Spanish-Columbian corpus are currently in preparation; we invite contributions of other corpora.
As part of the programmable corpora concept, EcoCor comes with so-called microservices: computational functions that algorithmically perform annotations or calculations on the texts in the corpora against the background of ecocritical research. For the EcoCor prototype, we have implemented an initial experimental pipeline that annotates entities of the type ‘animal’ and ‘plant’ from the texts based on an entity dictionary obtained from Wikidata. These entities are then available for further processing in the EcoCor frontend and in the EcoCor API. With this initial microservice, we refer to established works from Animal and Plant Studies and tie in with current DH research into biodiversity in literary texts (Langer et al. 2021 and 2024; Piper 2022) while also drawing on the fundamental idea behind EcoCor, i.e., using the platform to foreground genuinely ecocritical perspectives on literary texts, especially the multifaceted depiction of ecological aspects as part of a network structure in texts and corpora. Currently, we are exploring ways to complement microservices with machine learning techniques of entity annotation that ensure both stability and flexibility in their application to digital ecocritical research (cf. Schumacher et al. 2025, Piper 2024). One approach involves storing various annotation layers in the EcoCor database.
As already pointed out, the goal of EcoCor is to make ecocritically remarkable literary texts visible, to promote ecocritical perspectives on texts, and to foster ecocritical research by opening up new research questions. The EcoCor frontend, which is based on the reused DraCor frontend, is our central interface for human-machine interaction. However, the frontend does not only function as a corpus browser and full text reader (Fig. 1; Fig. 3). Through the currently implemented integration of entity clouds (on corpus level, Fig. 1; on single text level, Fig. 2), it also refracts traditional perspectives by algorithmically decomposing the text into a weighted cloud of all ‘non-human‘ entities that are populating it.
Via the EcoCor API, we open up the EcoCor data for machine interaction and thus also for computational analysis. In addition to general metadata on the corpora and individual texts, we currently use the API to output data from the entity annotation microservice.
These data can be fed directly into downstream computational analysis. For a showcase study, we processed the data provided by the ‘/corpora/{corpusname}/texts/{textname}/entities‘ endpoint with a Jupyter notebook and visualised it, using a python script written together with Claude Code, as a network (Fig. 4). This network shows which ‘more-than-human‘ entities (“animals”, “plants”) are – based on a window of 1 paragraph – co-mentioned in Emily Brontë's “Wuthering Heights“ (1847), leading to the novel becoming readable as, for example, a narrative about the companionship between dog and horse.
The API also makes it easy to retrieve entity information for entire corpora. This opens up the possibility of merging the networks from individual texts and creating a visualization of the overall network of non-human life in the texts of the corpus (see Fig. 5 for EcoCor-DE), providing e.g. an analytical illustration of the central role of the tree (“Baum”) in human storytelling.
At the 2026 Hackathon, we also developed a way to provide various entity annotations via the API, enabling the system to process even more complex and potentially conflicting annotation layers based on different conceptualizations of entities. Following the code review, this annotation endpoint will be implemented in the coming months.
EcoCor is conceptualized as an evolving and ever-growing corpus project with the aim to contribute to Digital Ecocriticism by providing a new infrastructure and further enabling a variety of interdisciplinary digital analyses. Currently, our main focus is collaborative corpus development. We plan to further expand EcoCor-EN and EcoCor-DE and to add new corpora. Alongside this, we are upgrading the dictionary-based entity annotation microservice with an individually trained machine learning classifier that automatically annotates references to animals and plants beyond biological taxa, capturing the specificities of narrative ecologies (Schumacher et al. 2025) and enabling us to identify ambiguities as well as metaphorical usages. We are also adapting DraCor’s network calculation microservice so that the analysis we showcased in section 2.4 can be carried out within the EcoCor platform, with results accessible via both frontend and API.
As EcoCor can be understood as social software operating in the mode of “sympoiesis” (Haraway 2016), these steps are embedded in ongoing discussions within a growing research collective, in which we also recognise the tension between computational abstraction and formalism on the one side and ecological specificity and materiality on the other. We consider this collectivist dimension crucial. Ultimately, following Latour’s rethinking of the social, EcoCor aims to serve as a modest infrastructural component within a dynamic “assemblage” (Latour 2005) committed to the ecological dimensions of humanities scholarship.