PALimpsests: Artificial Intelligence applied to advanced imaging of recycled manuscripts from Northern Italy in the Early Middle Ages, to study their production and use within a sustainable ecology

Projektbeschreibung

Neue Technologie zur Enthüllung versteckter Texte

Unter jahrhundertealter, verblasster Tinte verbergen sich verlorene Stimmen der Vergangenheit. Im frühmittelalterlichen Norditalien wurden alte Manuskripte abgewaschen und wiederverwendet, wodurch Palimpseste entstanden – ausgeblendete Schichten der Geschichte, die darauf warteten, enthüllt zu werden. Diese wiederverwendeten Texte enthalten Hinweise auf die kulturelle Wiedergeburt, die die karolingische Renaissance auslöste und die Kluft zwischen der klassischen Antike und der mittelalterlichen Erneuerung verringerte. Vor diesem Hintergrund setzte sich das ERC-finanzierte Projekt PALAI zum Ziel, diese Geheimnisse mithilfe leistungsstarker Bildgebungsverfahren und KI zu lüften. Durch die Kombination von multispektralen Scans mit maschinellem Lernen wird PALAI die gelöschten Schriften aufdecken können – schneller und deutlicher als je zuvor. Dieses Werk verspricht nicht nur, vergessenen Texten neues Leben einzuhauchen. Es wird auch unser Verständnis davon erneuern, wie Schriftgelehrte das Wissen formten.

Ziel

PALAI will study palimpsests using advanced imaging and artificial intelligence. Palimpsests, or recycled manuscripts, especially those from Northern Italy in the early Middle Ages, are known as a treasure trove of lost texts. The reuse of ancient manuscripts shaped the scene upon which the Carolingian renaissance emerged, and is thus a missing link between classical Antiquity and the first period of major cultural renewal in the Middle Ages throughout the West. PALAI proposes to investigate where the recycled manuscripts came from, how they were selected and prepared for reuse, and what their reuse tells us about scribal communities’ worldviews. Doing so requires multispectral and XRF imaging, to recover the erased features of the original manuscripts. A major issue with such imaging is the post-capture data processing, which is both excessively time-consuming and often fails to extract the relevant data for scholars to examine: crucially, PALAI proposes the first large-scale exploration of machine learning models in order to solve this image processing bottleneck issue, which will have numerous applications to erased or damaged documents from other areas and periods. Using these technologies, PALAI will track the production of palimpsests in Northern Italy from the late 6th to the early 9th century, by investigating, on the level of individual manuscripts, the cultural entities that fostered it; it will analyze scribes’ and readers’ intellectual practices, their relation to past cultures, to their present environment, and to innovation and creation for the future; and it will put forward in narrative form a multicausal theory of manuscript reuse bringing to light an ecology of writing in which long-term sustainability of texts was paramount. PALAI’s methodology is designed to combine approaches that are traditionally compartmentalized, such as manuscript studies, close reading and history, in order to foster interdisciplinary collaboration.