Katherine Faull, Moravian University
Moravians and Place: April 28, 2026
Moravian Archives talk: Sacred Confluence
The Scholar and the Source
The primary sources for this study are the Shamokin mission diaries, 1742-1755, which I translated and edited for publication as Cultures at the Susquehanna Confluence: The Diaries of the Moravian Mission to the Iroquois Confederacy, 1745-1755 (Penn State University Press, 2024), part of the Penn State Series on Anabaptist, Moravian and Pietist Studies. This prior work is methodologically foundational: I know this text as my tongue knows the back of my teeth. Every analytical decision described below was made against that intimate knowledge of the German original and its affective register, developed through over ten years of transcription and translation work.
Translation is, as Gayatri Spivak has argued, the most intimate act of reading. To translate the Shamokin diaries was to inhabit their emotional world: to feel the weight of the German word innig (intimate, heartfelt) in contexts of communion and companionship, and to register the shift to ängstlich (anxious, fearful) as colonial pressure mounted after 1753. That intimacy is not incidental to the methodology but rather foundational.
The broader historical and geographical context for the Shamokin mission is established in the monograph Cultures at the Susquehanna Confluence: The Diaries of the Moravian Mission to the Iroquois Confederacy, 1745-1755 (Penn State University Press, 2024). Also my deep research into the later missions on the North Branch, Friedenshütten and Scheschquehannunk (1765-1772), provides the comparative framework for understanding Shamokin’s distinctiveness as a mission place. My paper can be found here: Places of Peace: Moravian Missions on the North Branch of the Susquehanna 1769-1772
Corpus and Segmentation
The analytical corpus behind the paper “Sacred Confluence” consists of the Shamokin mission diaries from 1745 to 1755, the period for which diary records survive. The corpus comprises seventeen discrete sections, each written by a single missionary author, as follows:
| Section | Date Range | Author |
| 1 | September 13 – November 10, 1745 | Martin Mack |
| 2 | May 26 – June 28, 1747 | Johannes Hagen |
| 3 | June 29 – August 2, 1747 | Johannes Hagen |
| 4 | September 29 – December 31, 1747 | Martin Mack |
| 5 | January 4 – April 18, 1748 | Joseph Powell |
| 6 | April 18 – June 28, 1748 | Martin Mack |
| 7 | November 30, 1748 – January 31, 1749 | David Zeisberger |
| 8 | April 3 – July 26, 1749 | David Zeisberger |
| 9 | January 8 – March 5, 1750 | Christian Rauch |
| 10 | April 14 – June 2, 1753 | Bernhard Grube |
| 11 | June 4 – July 31, 1753 | Bernhard Grube |
| 12 | January 11 – July 2, 1754 | David Kliest |
| 13 | December 19 – 25, 1754 | Heinrich Frey and Gottlieb Roesch |
| 14 | April 1 – May 31, 1755 | Gottfried Rösler |
| 15 | June 1 – July 31, 1755 | Gottfried Rösler |
| 16 | August 1 – September 30, 1755 | Gottfried Rösler |
| 17 | September – October 1755 | Gottfried Rösler |
Each section was treated as a discrete unit of analysis, with equal weighting regardless of authorial identity. This decision was methodologically deliberate. Differential weighting based on prior judgments about individual diarists’ theological sophistication or narrative reliability would have introduced the scholar’s interpretive priors into the coding process before the analysis had begun. Equal treatment preserves the integrity of each authorial voice as a documentary witness in its own right.
The corpus has a structural feature that is itself analytically significant: the coverage is uneven. Significant gaps exist between July 1749 and January 1750, and between March 1750 and April 1753. The rupture year of 1755 is documented entirely by a single author, Gottfried Rösler, across four consecutive sections. The gaps in the record are not simply absences. They reflect the precarity of the mission’s documentary practice, the dependence of that practice on the physical presence of individual missionaries, and the increasing difficulty of sustained residence at Shamokin as colonial violence mounted. The silences speak.
The segmentation by author maps directly onto the argument. The emotional arc from Shikellamy’s years of companionship and protection (Mack, Hagen, Powell, 1745-1748), through the middle period of increasing pressure (Zeisberger, Rauch, Grube, Kliest, 1748-1754), to the rupture year dominated by Rösler’s single sustained voice (1755), emerges from the structure of the corpus itself rather than being imposed upon it retrospectively. Seven diarists across seventeen sections: not a single narrative but a succession of individual documentary witnesses, each bounded by the presence and departure of one missionary at the confluence.
Emotion Coding and the Plutchik Framework
The emotion coding was carried out using Robert Plutchik’s wheel of emotions as the governing taxonomy. Plutchik’s model organizes eight primary emotions (joy, trust, fear, surprise, sadness, disgust, anger, anticipation) into a graduated spectrum from intense to mild, with each primary emotion shading into adjacent states: ecstatic to joyful to content; frightened to anxious to uncomfortable, and so on.
This framework was not selected arbitrarily. I had applied Plutchik’s wheel in prior digital humanities work on the Moravian Lebenslauf corpus at the Moravian Lives project (moravianlives.org), a database of over 65,000 Moravian memoirs developed from 2014 onward. That earlier application established that Plutchik’s eight primary categories and their gradations map effectively onto the affective vocabulary of eighteenth-century Moravian German. The framework could accommodate the theological specificity of Moravian emotional language without flattening it. The Shamokin diary coding therefore built on a tested and calibrated instrument rather than starting from scratch.
Human-AI Collaborative Coding
Emotion identification and tagging was carried out collaboratively using Claude (Anthropic, claude.ai, January 2026), a large language model, to identify candidate passages in the diary text where emotion language appeared in proximity to a named place or named person. Each candidate passage was then verified against my own reading of the German original. The philological judgment was mine throughout. Claude operationalized categories I had established; it did not generate them.
I want to be direct about what this means in practice. I brought to the collaboration an editor’s and translator’s knowledge of this specific corpus. Claude brought the capacity to work systematically across a large body of text and to surface patterns that close reading alone might miss. The risk in any AI-assisted coding of historical text is that the tool imposes anachronistic or culturally inappropriate categories on a source that requires specialist knowledge to read accurately. In this case, that risk was mitigated by my verification of every tagged passage against the German original, and by the prior calibration of the Plutchik framework against the Moravian memoir corpus.
This is not a claim that the method is perfect. It is a claim that the method is transparent. The decisions about what counts as fear, or trust, or moral injury in these diaries are interpretive decisions. They were made by a scholar who has spent decades reading eighteenth-century Moravian German, not by an algorithm. The algorithm helped me see the patterns. The interpretation of those patterns is mine.
This collaborative model reflects an emerging best practice in digital humanities work with historical manuscript corpora, one that this study seeks to model explicitly. As I have argued elsewhere in relation to the ethical challenges of AI-assisted work with colonial archives (HAVI grant proposal, 2026), the central question is not whether to use AI but how to ensure that the scholar’s interpretive authority governs the tool’s output, rather than the reverse. The Shamokin methodology is one answer to that question: human philological authority at every decision point, AI as a systematic assistant rather than an autonomous interpreter.
Data Structure and Network Construction
Each tagged passage produced a structured data point consisting of three elements: an emotion coded to Plutchik’s taxonomy, a referent (a named place or named person), and a temporal marker (the year of the diary entry). These triadic data points form the relational dataset from which the network visualizations were constructed.
Four visualizations were developed from this dataset. The Shamokin Diary Emotion Wheel presents the distribution of emotional states across the full corpus and by individual year, allowing direct comparison between the emotional register of 1747-1748 and 1755. The Smithy Network visualizes the people, emotions, and interactions associated with the blacksmith’s forge across the full period and by selected time ranges, showing the transformation of the forge from a space of companionship and productive work to a site of moral injury and trauma. The Susquehanna River Network performs the same analysis for the river, revealing its function as a morally neutral, theodical counterpoint to the ethical entanglements of the forge. The combined Shamokin Diary Network brings people, places, and emotional states together in a single interactive visualization, allowing the relational geography of the mission to be read as a whole.
All four visualizations are interactive, allowing users to filter by time period and by type of relationship or interaction.
Methodological Stakes
This methodology sits at the intersection of philology, translation studies, and digital humanities. It resists two common reductions: the computational approach that processes historical text without philological grounding, and the close reading approach that cannot scale across a multi-year, multi-author manuscript corpus. The combination of Plutchik’s tested taxonomy, annual segmentation, human-AI collaborative coding, and network visualization allows the emotional geography of the Shamokin mission to be reconstructed at a level of granularity that neither approach could achieve alone.
The method also makes a claim about evidence. Emotion language in the Shamokin diaries is not incidental. It is the primary medium through which the diarists processed their experience of place, person, and divine action. To read the diaries without attending to their affective register is to miss their argument. Therefore, a placeography and personography of emotions are not supplementary to the historical analysis. They are its substance.
Finally, this methodology makes a claim about responsibility. To use AI to assist in the analysis of colonial archives is to enter ethically complex territory. The archives themselves are the products of asymmetrical power: European missionaries documenting Indigenous lives, in a language those lives were not lived in, for an institutional readership those lives were not addressed to. To add another layer of algorithmic processing to that already fraught record requires care, transparency, and the constant reassertion of the scholar’s interpretive authority. The method described here is one attempt to exercise that care. But there is much work still to be done.
Moravian Studies and Digital Thinking and Praxis 