Between Miracles and Memory

Min(d)ing the gap in construction of authority in early medieval episcopal saint’s lives and deeds of bishops

A mother finds her unborn son’s name spelled out in ceramic tile buried underneath a centuries-old church graveyard. A greedy count chokes to death on his juicy steak because he chose dinner over a saint’s request to return a cow to its impoverished owner. The holy vision of a bishop’s shining white vestments turn the knives of would-be assassins.

Each of these miracle stories are unique in their own way, but they share in the common tropes of a saint’s life: early childhood blessing, mid-life struggle against the stream of world-bound contemporaries, and a visionary death, all contributing the memorialization of a person literally larger than life. With the commemoration of these miracles by saints in the past, authors yoked a remote and mythological divine authority to the contemporary authority of priests and monks whose role as judge, shepherd, provider and protector extended beyond the sacral realm into the physical one. In turn, the physical connection between individual protector and protected–a bishop’s see, an abbot’s monastery, a saint’s patron city–contributed to the formation of a collective identity drawn from memory as much as present experience.1

For a variety of reasons, this is at least a partial return to something I looked at closely in my dissertation: the role of memory as it contributes to the propagation of a community’s reputation (and therefore it’s well being), whether that community is secular or regular. The difference in this current foray into memory is twofold: first, I focused on the differences in the construction of these memories for individual saints whose miracles appear in a dedicated vita and for bishops whose lives form part of an episcopal lineage contributing to a gesta episcoporum. Second, this is the first foray I’ve made into one particular kind of digital tool: topic modeling, although I’ll use several tools today to give the topic modeling more context. As such, it’s on a limited corpus of texts, only 475000 words distributed across 17 texts, which was built using the system described here:

As I worked through these texts, I expected to find memory building at the center of both the vitae and the gesta. In fact, memory, memoria, or the act of remembering, memoro, are neither extraordinarily common or explicitly tied in linguistic terms to audience action in either genre. In this visualization of how word placement works in this corpus, memoro is pushed to the outer right hand corner, the small pea-green offshoot that’s somewhat distant in linguistic terms from the rest of the word cluster. In other words, audiences aren’t explicitly being exhorted to “do this in memory of me”,3 an important note given the idea of a gesta episcoporum or gesta abbatum as a prescriptive behavioral text meant to bring the ghosts of bishops past into the day-to-day lives of actual bishops present and future.3 Instead, the language of gesta and vitae are dominated by individuals and institutions. Memory is implicit in the language of these stories, and in their introductions and conclusions, but we are not reminded regularly that memory is at the center of these texts.

The two genres share their focus on implicit, rather than explicit, memory building, but the affective focus of the memories built in each genre are very different. Gesta authors build an affective emotional connection for their audiences with an individual and bring that connection to bear on a positive outbound message they are crafting about their community and its role in a larger world. Vitae, on the other hand are inward facing, linguistically dominated by a specific community and only then by individual lives lived in the service of (and the walls of) their community.

In both cases, the memory formation happens at the moment when individual translates to communal.4 The audience’s affection is directed toward the community that accepts a saint’s holiness, rather than the saint. It’s no surprise that vitae focus inward in order to, in vulgar terms, market outward. What I was surprised by is the role community building has in the palatability of each saint. Vitae build up the reputation of a community by showing the outside world how the community’s saints blossomed only once the saint was united with that community.5 Gesta authors on the other hand build up the individual bishop prior to his entry into the community. Given that thie historiographic read on gesta is that they construct internal lineages to make the history of the institution stronger so individuals in the present can benefit, I was actually surprised. Instead, gesta bring individual authority to bear in support of the institution to which those individuals contribute. To make an oversimplified modern analogy, vitae are 1940s vacuum cleaner ads–“you won’t be happy until you’re one of us”–while gesta are WWII Uncle Sam recruitment posters in the vein of individual responsibility: “fight on our behalf to make ‘us’ better”.

Individual word choice in gesta and vitae

Let’s take a look at how this works by starting with an overview of the documents and what I’ve done to them. For the early middle ages, between the seventh century and the ninth century, I’m working with 7 sources. Three gesta contribute 62,000 words and 4 vitae contribute 45,000 words. For the central Middle Ages, between the tenth and the twelfth century, you’re looking at information from 9 sources, with 2 gesta contributing 173,000 words and 7 vitae contributing 105,000 words.

The first piece of valuable information these basic numbers provide is that gesta are slightly more diverse in their vocabulary. Over all, the gesta contribute about 235,000 total words to the corpus consisting of 10,000 unique words in the text with the vitae contributing 150,000 words total and 8,500 unique words. If the two corpora are processed together, there are about 11500 unique words, suggesting that the variety of language in the two corpora–the number of unique words in gesta vs vitae–is about even. NB: This kind of word analysis can work on corpora as small as 5000 words, but it’s best to have more and clearly we’re well above the minimum 5000 word threshold.

One of the persistent claims about medieval hagiography is that that vitae share so many tropes across lives that they’re overly repetitive.6 The comparative variety of language across these two corpora, particularly given that when the two corpora are added together, there’s an additional jump in unique words, suggests that’s not an accurate claim. From a historian’s perspective, that means our vitae authors aren’t dependent on remembered tropes of saintly behavior to communicate the value of an individual saint to their audience, or at least not on the repetitive language that scholars tend to ascribe to the repetition of saintly tropes.

Topics in gesta and vitae

Early medieval (7-9th centuries)
  1. ecclesia ipse hic imperator voco unus ubi pars infra usque
  2. is se idem episcopus cum nec de habeo magnus veroverus
  3. qui video ex iohan civis per corpus urbs virvirumvirusvis verbum
  4. ego dico tu hic tuus et meus ille sum quia
  5. sum qui rex cum civitas veroverus super egredior expleo induo
  6. et ad post pontifex cum dies ex mitto multus praedico
  7. et in omnis sum se dies autem homo contra terra
  8. et non qui sed dominus vos deus vester possum sicut
  9. ut suus tantus atque hic ita vir etiam semper unus
  10. sanctus in qui beatus sum super invenio apolen aedifico sub
  11. sum dico nos audio quia cum talis veroverus manus ecce
  12. non in suus de ut iste sum pro per sed
  13. is ille et alius tunc suus ut venioveneo omnis ad
  14. annus de corpus autem vita suus tempus sepelio mensis sedeo
  15. et sum ravenna tempus iste ab rauenn nunc aureus teneo
  16. per christus pater filius sed et terra duo benedictus alius
  17. que sanctus omnis deus multus quum sed dum cunctus ac
  18. qui in et monasterium ad beatus vel frater non gratia
  19. qui is ad in ille dominus video ut ipse jam
  20. ecclesia ipse hic omnis ex sum et sedes papapapas unus
Central medieval (10-12th centuries)
  1. ipse se hic in video corpus beatus de divinus tantus
  2. hic is ecclesia idem se ut etiam ab super quoque
  3. et in a nos noster sum dico vos non ille
  4. remigium cum qui praesul rex ac remetior remus suus post
  5. ipse pro de idem item quidam is rex res se
  6. et hic ad habeo sum qui regnum per ipse rex
  7. qui episcopus et ad is in sum papapapas cum quidam
  8. in is sum unus faciofio magnus vita pars inter qui
  9. rex qui non ut sed cum enim multus pro ne
  10. in sanctus ecclesia qui suus quoque de res idem a
  11. et qui meus noster ecclesia cum sancio ex episcopus nomen
  12. qui sanctus quidam ad hic cum nomen ubi beatus dum
  13. et qui ut suus non ille in is si de
  14. qui ille et sum omnis nec dico si de tempus
  15. et vester sum auctoritas ac scribo ad alius de sedes
  16. archiepiscopus in sed cum a sedes ab ecclesia post pontifex
  17. in tu ego sum is et tuus qui dominus dico
  18. suus et is ut se vel ex episcopatus venioveneo deus
  19. urbs ab sui tandem pater virtus jam lumen salus vir
  20. sum et qui monasterium monachus ibi post hic deus annus

It’s no shock that “ecclesia” or church is high in both periods’ topic lists. That’s the equivalent of doing a text mining exercise on the bible and finding out it’s about Jesus. From there, though, we find topics that contain words describing individual roles and their various characteristics in relationship to the community at large. The yoking of emperor to early medieval church, but as one called voco to be part of the church pars tells us how our early medieval gesta authors want us to remember the church’s authority in these early years. The church is still high in the central medieval topic list at number two–here is the church, hic ecclesia, which is above everything, super quoque.

The next few topics in each list, though, are dominated by individuals. The first topic in the central medieval gesta is oriented toward seeing the most divine, tantus divinus, bodies of individual saints, beatus corpus. It’s similarly revealing that the bishop, episcopus in the early medieval topic 2, is possessed of, habeo, great truth, magnus verus. Existing analysis of gesta episcoporum support the focus we see here on the individual by noting a heavy dependence on specific people whose protection shores up the community.7 In topic 4 in the early medieval gesta, we see a bishop named John, whose body is the of the city itself. In the central medieval gesta we see Remigius in topic 4, who is protected by, praesul, (or maybe protecting) the king, rex.

Individual kings also play a role in embodying the landscape of the bishop saint. We’ve seen Remigius and his protective relationship with rex. In early medieval topic 5, it’s slightly less positive: a king, rex, clothes himself, induo, satiates or fills himself up, expleo, before leaving, egredior. This shows us early medieval gesta authors writing in memory of bishop-saints focus on the city at the center of a bishop’s see and the depredations it suffered at the hands of greedy king.8

The traditions of the cult of saints also build up the bishop’s environment in the language of a gesta. Saintly corpses buried for years are treated with language that also describes the length of a bishop’s tenure, tying the authority of these individual bishops to a lineage of miracle workers. These miracle workers benefit the church at large, rather than smaller institutions inside the church (like episcopatus or the episcopacy). And at long last, monasterium finally crops up to demonstrate the degree to which individual reputation is leveraged to personify the community to which that individual belongs.

The expression of this public role, plays itself out in larger patterns of episcopal relations with landowners and local clergy. in other words, bishops were intent on navigating the boundaries of conflict with other authority figures, and with the local urban population, who themselves had a stake in local and regional administrative duties in the wake of now-absent Roman power.

Missing? Memory and commemoration. These memorializing words, though they feel frequent in a read of vitae and gesta, actually appear both less frequently, and less consistently with other frequently appearing words, that from a linguistic vocabulary standpoint they become almost meaningless. In other words, memory isn’t highly associated with a single overarching concept: not vir, not beatus, not sanctus, not ecclesiae. Consider that in the context of the fact that episcopal names show up on this topic list (Iohannes), and you can see both the strength and weakness of an analysis like this: you can readily get at topics that show up frequently, but an overarching concept that plays heavily in an introduction or conclusion may not dominate the body of a text enough to be readily apparent in a topic model. We’ll get to that in a minute.

Early medieval (7-9th centuries)
  1. et in ille domus possum qui ubi locumlocus manus mors
  2. ad qui deus facio venioveneo ibidem quidam iter suscipio usque
  3. et noster ad inter coelestis beatus nam virvirumvirus pater unde
  4. sanctus hic verbum aio tempus aliquis valde alius a audio
  5. suus ac omnis corpus per enim nec atque quia episcopus
  6. et sum ipse ecclesia felix vel divinus gratia quaero qui
  7. sum se habeo martin video ex non monasterium quis jam
  8. in sum is ille dominus suus omnis tuus jubeo apud
  9. sed qui non et a tantus etiam ut ne quasi
  10. de ad cum qui in virvirumvirusvis aqua a rex cito
  11. qui sum ut cum nam ita se vox nomen statim
  12. qui non vita ut si aut ipse unus vel nullus
  13. cum tum veroverus tempus is turba sto corpus signum ut
  14. idem in sanctus post quidam hora nunc columba nomen frater
  15. is se ipse multus frater ab autem oro oratio refero
  16. qui is et in vir autem coepio invenio intro gaudeo
  17. hic de dico ad video monasterium annus terra christus ante
  18. in qui dies nox vir angelus suus anima coelumcoelus oculus
  19. sum ego ille inquiam is quia dominus meus ut nos
  20. et dies per deus duo contra nos omnis unus iuxtajuxta
Central medieval (10-12th centuries)
  1. qui sum in dico et ut habeo si sed non
  2. deus qui et meus sanctus video spiritus vox talis divinus
  3. in se et is autem monasterium romual possum venerabilis alius
  4. ille cum quidam ut nec sed per aut do ita
  5. qui sum ille is video ipse ad veroverus quidem jam
  6. et sum in ipse deus omnis facio noster idem hic
  7. qui se ut ad geral tamen sum possum soleo nam
  8. hic per nam virvirumvirusvis tunc beatus tantus tam nos sic
  9. qui ad suus ecclesia ab lanfranc res card scribo vita
  10. in et is suus omnis non tempus ab super ubi
  11. vel tu tuus ille dominus quia ne non quis nomen
  12. et ut ille quidam idem venioveneo filius a locumlocus miles
  13. sum hic qui ego dico non in enim vos sicut
  14. et abbas ac in ut tantus atque annus omnis locumlocus
  15. is qui suus pro autem veroverus hic possum cum manus
  16. jam frater vir per virtus dum ita corpus mens vis
  17. is episcopus cum ad in sum et coepio se civitas
  18. non ad ex sed ut quia cum ipse faciofio etiam
  19. de et qui is sanctus rex a sum ex petrus
  20. et cum dies ad sanctus deus ecclesia corpus alius ibi

For the vitae in the project, the story changes. It’s immediately apparent that the issue of community can be bad–as in topic 1 for the early medieval vitae, In that building was the hand of death–or good, in topic 6, divine grace grants happiness to the church. References that tie an individual to a specific role in the institution, like bishop or pope or king or emperor, are missing in favor of the fairly generic sanctus–holy man–and the very humble vir, the medieval equivalent of “that guy”, though in this case our humble vir are marked by their service to pater deus, our father in heaven. Only 5 topics down down do we see our first mention of a specific institutional role, episcopus.

As with the gesta there are some changes that reflect (and in many cases are designed to affect) the contextual changes that mark the multitude of changes in monastic life between the 10th and 12th centuries. We see an individual name, Romuald, placed highly in the central list in topic 3 but his role as an individual figure of holiness is made possible by his integration into the institution: Romuald can finally be a venerable figure venerabilis for others alius only within the walls of monasteries monasterium. Keep in mind, though, that Romuald appearing here as an individual means the other words hew closely together in saints lives other than Romuald’s, it’s just that Peter Damian associated Romuald’s name more closely with these words than other authors associated their saints’ names.

Contextualizing memory and miracle

These topic-model word buckets and the overview of word usage that goes with them, are very handy analytical tools, but they need context. Here, a look at corpus linguistics and network analysis is helpful. If you iterate through a document, word by word, and grab the words around it in pairs, rather than word by word, you get a more precise set of relationships between words. With each of those word pairs represented in a network diagram, our early medieval corpus, both gesta and vitae, looks like this.

This particular layout uses the strength of the relationship, or edge in network terminology, between words or nodes, to adjust spacing and placement in the diagram. Words that show up a lot and are connected to other words that show up a lot are in the center, and we can filter a bit to get a better look at the concepts we’re interested in.

Next, I filtered the original blob for some of the words that have come up here: bishop, episcopacy, church, monastery, abbot, memory. The filtered set of words aren’t distributed randomly throughout this big blog. Instead, they’re closely related, a blobby cluster in the center of the original big blob.

The "ego" filter here looks for words that appeared in the topic model (somewhat narrowed by my analytical choices), and then displays 1 level of direct node-to-node connections so we can see how the words in the topic model are connected and to what other words they are connected.

Expanded a little for legibility, the filtered portion of our network looks like this:

The filtered set of words aren’t distributed randomly throughout this big blog. Instead, they’re closely related, a blobby cluster in the center of the original big blob.

Episcopus, sanctus and vir all sit in the same large “community” of related nodes, the blue on the left, while ecclesia, monasterium and construo (build) all sit in the green community that dominates the right of the network, along with other words like relinquo that suggest sanctified episcopal control of the monastic community on the right. In a corpus-linguistics analysis of the texts, sanctus and episcopus or sanctus and a bishop’s name are linked (sanctus appearing either directly to the right or the left of the other term) more than 40 times. Abbas, abbot, is only linked to sanctus 4 times, with beatus at similarly skewed ratios. Consider that this part of the corpus includes a gesta abbatum, or deeds of abbots, and the idea of the outward facing emphasis in gesta, whether for secular or regular clergy, begins to dominate.

It’s similarly in this integration of network analysis, corpus linguistics and topic modeling that we begin to understand memory and its distribution. Memoria the noun is in the brown cluster of words related to veneration, venero, and recollection, recordatio. It’s node is fairly small, which means it doesn’t appear very often or have many connections other than to the words in this small cluster, but there are two that are important: contrado and causa, or “delivered together” and “purpose”. Neither are directly connected to memoria, but they are clustered near it, which means the connections between contrado, causa, and memoria to other words are similar. The delivery of memory with a purpose–and here we’ll look at memoriale closely connected to sempiterno in the lower right corner of the visualization, everlasting memorial–is to deliver an everlasting memorial. Distribution is exactly the key. The implicit movement of memoria as a concept throughout these texts, never overwhelming the institutions or the individuals who need to be memorialized, and the close connection between memory and wide distribution throughout all time is quite nicely represented as a gap, as its absence from the topic models.

The secular outward facing nature of a gesta and its insistence on the individuals constituting the whole and defending the community against outside depradation supports some of the more recent historiography on gesta, which suggest that attempts at royal control of monastic and episcopal holdings created serious economic issues for religious institutions.9 This emphasis was fomented by the expansion of Merovingian and Visigothic power and cemented by the Carolingians, rather than being purely a Carolingian phenomenon. Vitae, on the other hand, focused on attracting visitors and donations, rather than on defending historical precedent for ownership and authority. While the two genres are often treated together because miracle content and biography ties them together, the divergent purposes, highlighted here by considerable linguistic difference, suggests both the value of an approach like this and the importance of including serial biography in order to understand the memorialization of saintly authority.

  1. Rosamond McKitterick, Rosamond McKitterick, History and Memory in the Carolingian World (Cambridge: Cambridge University Press, 2004), 282; see also James Fentress and Chris Wickham, Social Memory: New Perspectives on the Past (Cambridge, MA: Blackwell, 1992). 

  2. For more on lemmatizing and the use of lemma in textual analysis, see Arne Fitschen & Piklu Gupta, “Lemmatising and morphological tagging,” in Anke Lüdeling & Merja Kytö (eds.), Corpus linguistics: an international handbook, Vol. 1, 552-564 (Berlin & New York: Walter de Gruyter, 2008); Sali Tagliamonte, “Representing real language: consistency, trade-offs, and thinking ahead!” in Joan C. Beal, Karen P. Corrigan, & Hermann L. Moisl (eds.), Creating and digitizing language corpora. Vol. 1: Synchronic databases, 205-240 (Houndmills: Palgrave Macmillan, 2007); Gregor Wiedemann, “Opening up to Big Data: Computer-Assisted Analysis of Textual Data in Social Sciences” in Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, Vol. 14 No. 2 (25 May 2013). 

  3. Michel Sot, Un Historien Et Son Eglise Au Xe Siècle: Flodoard De Reims (Fayard, 1993), 22-3.  2

  4. Constance Bouchard, Rewriting Saints and Ancestors: Memory and Forgetting in France (University of Pennsylvania Press: 2014), 49, notes that the individual affect was often a result of the author having a personal connection to his subject. See also Fentress and Wickham, Social Memory (Cambridge, MA: Blackwell, 1992), 45-49. 

  5. Athanasius’ Life of Antony provides the model for this. See also Aviad M. Kleinberg, Flesh Made Word: Saints’ Stories and the Western Imagination (Harvard, 2008; trans. Jane Marie Todd from Histories de saints: Leur role dans la formation e l’Occident, Gallimard, Paris, 2005). 

  6. Emma Campbell, Medieval Saints’ Lives: The Gift, Kinship and Community in Old French Hagiography (Cambridge: Boydell & Brewer, 2008), describes this phenomenon–and some of the work debunking it–on pp 6-8. Her argument is that hagiography isn’t linguistically original but that it is thematically and ideologically original. I argue here that it is at least not linguistically repetitive to the degree historiography suggests. 

  7. Michel Sot, Gesta Episcoporum, Gesta Abbatum, Typologie Des Sources Du Moyen Âge Occidental (Turnhout, Belgium: Brepols, 1981), 32-3; Deborah Mauskopf Deliyannis, “A Biblical Model for Serial Biography: The Books of Kings and the Roman Liber Pontificalis,” Revue bénédictine 107 (1997): 16; McKitterick, History and Memory in the Carolingian World, 2-4. 

  8. Joaquín Martínez Pizarro, Writing Ravenna: The *Liber Pontificalis of Andreas Agnellus, *Recentiores (Ann Arbor: University of Michigan Press, 1995), 20; Agnellus, The Book of Pontiffs of the Church of Ravenna, trans. Deborah Mauskopf Deliyannis, Medieval Texts in Translation (Washington, DC: Catholic University of America Press, 2004), Introduction, 77-80; Paul, Vitas Sanctorum Patrum Emeretensium, trans. Antonio Maya Sánchez, vol. 116, Corpus Christianorum Series Latina (Turnhout: Brepols, 1992), V. Vi. 89-99. 

  9. Bouchard, Rewriting Saints, 138.