Work at the intersection of data visualization, literary studies and the spatial humanities has identified the shortcomings of digital cartographic tools and applications such as GIS in addressing the complexities and ambiguities of literary space. In this article I present a visualization approach which takes an alternative angle on this problem, using as an example the representation of scalar shifts in the poems of Emily Dickinson: the plotspace, a 3D-fabricated sculptural object which can be printed in plastic or resin or milled out of materials such as metal or wood. A plotspace incorporates lines of text engraved or embossed on its inner face(s), as well as a curved shape which, extending in a three-dimensional wave-like form, plots the shifts in the size of entities referred to in the text. Plotspaces, I argue, have the capacity to prompt and provoke their viewers to go beyond the usual processes involved in close reading, by considering what is illuminated by reading a text in conjunction with a three-dimensional form which fabrication technologies make it relatively easy for non-specialists to create for themselves. In addition to these literary and hermeneutic affordances, plotspaces also raise questions about what it means to model spatiality in literary texts, and what tactility, materiality and sensory apprehension might add to the interpretive experience of reading a poem. They also speak to prevailing concerns engaging feminist visualization scholars about the conditions in which knowledge objects such as data visualizations are produced, and the shaping effects of those conditions.
This article introduces True Digital Poiesis (TDP), a conceptual pedagogical framework and proposed interface design for engaging generative AI in the Humanities. Drawing on experiential learning theory and grounded in the humanistic preoccupation with alterity — the otherness of history — TDP proposes the use of AI to simulate premodern events while foregrounding the active interpretive effort involved in constructing the past. The framework envisions the design of a digital prompt-builder, the Event Cruncher Stylus (ECS), and a structured critical review process, both of which are designed to support student engagement with the complexities of historical construction. In presenting the theoretical foundation, practical applications, and interface design of TDP, this article argues that precise and responsible use of AI in historical simulation — ranging from non-immersive to fully immersive simulative modes — requires a systematic, embodied method for engaging the alterity of the past. As a conceptual proposal grounded in hermeneutic theory and experiential learning research, TDP is presented here as a framework whose empirical evaluation through classroom implementation and outcome assessment is a primary objective of ongoing and future research.
Topic models are highly sensitive to corpus construction, preprocessing, stopword removal, segmentation, model implementation, and the selected number of topics. Yet rather than present these limitations as something new, this article revisits topic modeling to ask what standards of transparency and reproducibility should follow from them. Through a case study of Fernando Pessoa’s Livro do Desassossego (The Book of Disquiet), we compare topic modeling outputs across different editorial versions, preprocessing conditions, and modeling configurations. Using Quanteda for corpus preparation and document–feature matrix construction, and seededlda for unsupervised and guided topic modeling, we show how apparently technical decisions shape outcomes. The article argues that topic modeling should be understood as an exploratory interpretive procedure rather than as a neutral mechanism for discovering themes. Its scholarly value depends on making the full modeling pipeline — from source acquisition and cleaning to parameter selection, code, derived data, and interpretation — available for inspection, reuse, and critical contestation.
The increasing reliance on open-source datasets for training large language models has revealed a critical oversight in artificial intelligence development: the presence of historical toxicity embedded within canonical literary texts. This article examines the application of contemporary toxicity detection models to the Chanson de Roland, one of medieval French’s foundational literary works, exposing significant challenges in identifying hate speech, violence advocacy, and discriminatory content in historical documents. Using the multilingual Detoxify model on Joseph Bédier's modern French translation, I analyse 2,605 sentences to assess how toxicity detection models trained primarily on contemporary social media content perform when evaluating medieval literature. My findings reveal a troubling pattern: while the model successfully flags some explicit threats and insults, it systematically fails to detect the text's most problematic content, including religious misrepresentation, forced conversion narratives, and anti-Black racism. These false negatives represent a fundamental problem for AI models being trained on historical open data, as they risk perpetuating and amplifying centuries-old discriminatory frameworks while appearing objective. This research contributes to urgent debates about responsible AI development, arguing that without human-annotated ground truth datasets specifically designed for historical texts toxicity, current LLMs risk distorting historical understanding and diffusing undetected violent and racist discrimination at unprecedented scale. I advocate for interdisciplinary collaboration between computer scientists and humanities scholars to develop ethical frameworks for curating historical datasets that acknowledge their toxic content without either sanitizing history or amplifying historical harm.
This case study presents challenges and solutions for building calibrated syntactically annotated diachronic corpora of French. Through a series of integrated projects that developed methods for syntactic annotation of medium and large collections of non-literary texts over a long period of time (13th – 19th centuries), we consider how the availability of tools and the emergence of research questions shape and direct corpus-building activity. Supported by the experience of an earlier project, a semi-automated workflow for syntactic annotation was developed to allow processing of two corpora calibrated by provenance and genre, with a focus on parts-of-speech tagging in three annotation frameworks. The workflow combines the use of a neural dependency parser, a lexicon of lemmata and forms and a Python library. This experience, in turn, gave rise to a new initiative that aimed at formulating and testing best practices for creating syntactically annotated treebanks for different periods of the history of the language.
