UC Berkeley

For years, scholars of medieval German have grappled with how to analyze formal characteristics of the lyric and epic poetry while taking into consideration performance, musicality, mouvance, linguistic variation, and aggressive editing practices. The scholarship has justifiably resorted to restricted explorations of specific texts or poets, or heavily criticized region-specific descriptions with several caveats. The many challenges this multifaceted poetry presents has also obscured one of its most central features—–the medieval voice. With the little evidence we have often being ambiguous or contradictory, how are we to understand the role of the medieval voice in the German corpus as a whole? This project seeks to shed light on this forgotten aspect by taking advantage of computational methods to demonstrate

relative formal and thematic relationships based on sound and voice. In doing so, it presents several new prosodic analytical methods.

The first chapter of this project underlines the importance of sound and voice to medieval performance and composition. It additionally justifies the syllable as the foundation of the novel formal methods presented in the following chapters. Chapter two presents a new

syllabification algorithm that combines two linguistic principles: the Sonority Sequencing Principle and the Legality Principle. This algorithm is then customized and computationally implemented to perform accurately on medieval German across dialects and editing practices. Chapter three employs this syllabification algorithm to characterize different phonological soundscapes in reference to theme and voice using a modest corpus of only medieval German lyric poetry. While chapter three intends to quantify the vocal affect of a soundscape, chapter four aims to account for the sequencing of these soundscapes within the larger text and corpus. Chapter three therefore lays the foundation for a structural model of medieval German form. Finally, chapter five reduces the corpus once again to a very small subset of medieval German Vierheber (epic poetry with four stresses per line). It presents a supervised machine learning model to predict scansion for these texts according to the theory proposed by Andreas Heusler, and draws conclusions from how different poets took advantage of the freedom this structure allowed.

In each chapter, I present the aggregate statistics to confirm and supplement our knowledge of the medieval German corpus as a whole. Yet more importantly, I return to individual texts in order to demonstrate how these newly discovered formal soundscapes manifest and function within a smaller narrative. This combination of “distant reading” and “close reading” supports my overarching argument that medieval German form and content were demonstrably and quantifiably highly intertwined. In many cases markedly different formal, phonologically-influenced structures were implemented to trigger connections to related texts and formats. This project thus creates a new understanding of intertextual relations in medieval German literature.