Digital Tool-Assisted Hermeneutics: “Voyant Tools” and Medieval Hagiography

Salvete omnes! I am sorry that this blog is taking quite a while to get going. I am in the process of finishing up my course work for the term, and look forward to begin posting regularly once that is taken care of. I do, however, have a blog post to share today. This is a brief paper that I wrote for my class at Brock University with Dr. Michael Driedger, “Visualizing Historical Research.” It provides a quick look at some of the issues involved with using digital tool-assisted hermeneutical methods such as the online program “Voyant Tools” with medieval hagiographical texts. I hope, by the end, you will have at least a rudimentary sense of some of the areas of discussion we must turn to in the future as it relates to this topic. Here it is:

Hagiography, or saintly literature, is a literary genre perhaps most associated with the Middle Ages.[1] This is not wholly unwarranted. In the Latin west, for instance, the cult of saints was central to Christian piety from the patristic period until the Protestant Reformation.[2] Saints performed a number of roles in society; they functioned as powerful heavenly advocates, healers, social mediators, sources of cohesion, as well as symbols of political and religious identity.[3] One aspect of these cults is their associated hagiographic literature, which includes Latin and, increasingly from the twelfth century, vernacular Lives, or vitae. These works of sacred biography were products of their specific politico-cultural milieu; they served to consolidate as well as “tame” a cult, attempting to ground it in a clerically-defined religious discourse.[4] These texts, which for the Middle Ages are extant in the thousands, thus offer much to the historian interested in medieval notions of sanctity as well as issues of identity and everyday life. They are a window onto medieval mentalities.

The texts themselves have been the subject of serious scholarship since the seventeenth century with the Bollandists, whose Acta Sanctorum, published in 68 folio volumes between 1643 and 1794,is notable as an enormous collection of critical hagiography. Despite this long history, the study of hagiography has largely been neglected relative to other medieval literary genres due to confessional concerns, as well as, according to Manfred Gorlach, a preference among scholars for “high” literature.[5] Nonetheless, there have been some important contributions to the study of medieval saintly literature. Historian Andre Vauchez, for instance, has attempted to piece-together medieval mentalities in his 1981 book, Sainthood in the Later Middle Ages. Vauchez focuses primarily on the processes of canonization that were produced as official inquests into a saint’s life and cult between 1185 and 1421. He consciously does not center his study on saintly vitae. The vitae, he believes, due to their exclusively clerical provenance, are not ideal as indicators of popular mentalities. The processes of canonization, like inquisition records, immortalize the words of ordinary lay people. They are important as they reveal “what individuals who belonged to very diverse social and cultural milieux thought about a person they had known or heard talked about and what idea they had formed of Christian perfection.”[6] Vauchez considers a large group of these texts in order to produce conclusions – over seventy-one. He notes a fundamental difference between what he calls “mentality” and “mental structure.” Mentality, remained common to all, and was slow to change. Here, Vauchez points to the signs of sanctity, such as corporeal incorruptibility, that remained relatively constant throughout time.[7] Mental structure, by contrast, was changed and shaped by cultural trends. This explains the differences in interpretation of the signs of sanctity between the masses and the clerical authors of these texts.[8]

A more recent study is Eleni Tounta’s article, “Saints, rulers and communities in Southern Italy: the Vitae of the Italo-Greek saints (tenth to eleventh centuries) and their audiences” (2015) in the Journal of Medieval History. Tounta deals explicitly with saintly vitae. She conducts a close reading of these texts in order to gleam information about the relations between political authorities and local communities in medieval southern Italy. Her methodology is dependent on ideas of the interactions between text and community, as articulated by Julia M. H. Smith.[9] One of her conclusions, for instance, is that these vitae functioned to mediate between the Byzantine authorities and local communities.[10] Her work approaches the topic of hagiography with an eye to functionalist explanations for the cult of saints.[11]

This paper will draw from both Vauchez and Tounta. It will demonstrate both the possibilities and limitations of digital tool-assisted hermeneutics, as it pertains to medieval hagiographical texts. I would like to borrow from Vauchez’s commitment to mentality, although I am of the opinion that saintly vitae can and do indicate much more than exclusively clerical attitudes. This is where Tounta’s methodology will be useful. These texts were not produced in isolation. They were, in many cases, dependent on pre-existing oral traditions of lay provenance.[12] Thus, this paper will serve as a commentary on the capabilities of digital tool-assisted hermeneutics to contribute to a scholarly conversation about mentality as it appears in medieval saintly vitae. I will focus specifically on the web-based program Voyant Tools, a digital humanities project launched in 2003 that allows users to interpret a text or group of texts in a number of ways, as seen below. Here, my thinking has taken as its starting point Geoffrey Rockwell and Stefan Sinclair’s work Hermeneutica: Computer-Assisted Interpretation in the Humanities (2016). In fact, my interest in mentality has its counterpart in Rockwell and Sinclair’s discussion of “conceptual system” as expressed through words, an idea that has itself been borrowed from Roberto Busa (1980).[13] As the authors rightly point out, we must turn to an writer’s “verbal system” in order to understand, in a philologically-sensitive manner, his or her value system. The work of Rockwell and Sinclair also raises questions about the use of digital tools in hermeneutics and our conception of what constitutes a “text,” a discussion that as will be seen, holds a special importance for the medieval period. Lastly, I have chosen only the original Latin texts as part of my investigation, in the interests of remaining as true to the original as possible, but also in the hopes that I will be able to comment something on the relative strengths and weaknesses of working with Latin texts in general in Voyant Tools. Making no claims to be exhaustive, I hope that this investigation will raise some points for future discussion surrounding the use of digital tools in medieval studies and the humanities more broadly.

I began my inquiry simply by pasting a hyperlink into Voyant Tools of Fordham University’s “Internet Medieval Sourcebook” Latin vita of St Nicholas of Bari (d. 342), as it appears in Jacobus da Varagine’s (d. 1298) famous Legenda aurea. I decided to limit my experiments with Voyant Tools to the Legenda’s vitae, in order to help focus issues of authorship. After adjusting the tool to remove the pre-programmed Latin “stopwords,” “common” words like res (thing) and its corresponding declensional variants, I ended up with the following “cirrus” representation of the most frequent terms in the text, as seen in Figure 1:

Figure 1

Before proceeding, I would like to comment on the “stopwords” function of Voyant. Enabling this feature is not a necessary step in the analytical process. In fact, the decision to block all of the “common” words of any language, as they appear pre-programmed into Voyant, must first cause students to pause and consider what is being done. It is, at least on the surface, unclear who has decided which words should appear on the list of stopwords. There is, likewise, no immediate justification provided for the words chosen. The obvious concern here is that the tool may hide certain “common” words that are in fact analytically useful. The word res, for example, has a diverse set of possible “meanings” in addition to “thing,” like, “business,” “affair,” and “matter.”[14] The word is thus not necessarily always useful to block. But students must also be wary of the ideological choice that is inevitably made when words are ranked according to their perceived “value.” Users must interrogate the list of stopwords. Why are these specific words being blocked? What assumptions are being made? To what end? Perhaps it is safer to begin one’s analysis without any stopwords at all and to compare how the addition of the stopwords alters Voyant’s results. The “stopwords” function can be particularly problematic with a language like Latin, which is highly-inflected, and whose lexicon (like any language, to be sure) varies greatly depending on the historico-cultural context. A word and its corresponding “meaning” that appears in a Ciceronian text may have a very different connotation, for instance, a millennium and a half later in a saintly vita produced by an English monk. For the student interested in quickly placing a text into Voyant and immediately interpreting the results, the considerations outlined above are not necessarily a priority, however unduly. If the student has the time, he or she may be able to negotiate these issues by manually making their own list of stopwords.

The first thing that I noted about Voyant’s output for the vita of St Nicholas is that the word dei is used the most. One interesting aspect of the use of Latin (and other highly-declined languages) texts in Voyant is that it is possible to learn information about a word’s textual context simply by looking at the declension of the word in the “cirrus” view. This can provide clues, or indeed direct indicators, of the words that surround the word in question in the text, as well as the word’s grammatical role in a sentence. No referral to the “contexts” function in Voyant is necessarily needed. In this case, dei is the genitive form of deus (god), so its meaning is extremely likely to be “of god.” The fact that this term appears the most throughout the text (a total of thirteen times) is apparently initially unhelpful. This is a limitation of Voyant. The tool cannot substitute for close reading, although it can quickly pinpoint the desired sections of a text so that close reading may be conducted. In this case, a glance at each instance of the term in the actual text, as shown through Voyant’s “concordance” function, reveals that the term is mainly used as “servant” or “man” of god, and refers directly to St Nicholas himself. It is not surprising that a term used to refer to the vita’s “protagonist” should appear so many times in the text. There is, nonetheless, something of use here for our purposes. The precise title that is designated to a saint can reveal information about mentality. Here, one must tread carefully. Medieval hagiography is a literary genre whose language and form have taken their cues, to a significant degree, from formulaic literary topoi. The use of a typical designator such as, “servant of god,” is thus perhaps best viewed as simply just that. The interesting question is if and when there are divergences from the norm. Are certain saints referred to in unique terms? What does this reveal about the text, its author, and respective audience? Unfortunately, it is not possible to determine using Voyant the different ways that a saint is referred to in a text. The tool’s interpretive capabilities are far too general for such specificity.

I would also like to draw attention to the word iudaeus (Jew) in the vita of St Nicholas, which is the second most used term in the text, appearing nine times. What is a Jewish person doing in a text about a Christian saint? A look at Voyant’s “reader” function shows that the term is used exclusively toward the end of the narrative. Here, there is an anecdote about a fraudulent Jewish moneylender, who suffers an accident while traveling and becomes baptized as a Christian after receiving a miracle from St Nicholas. Varagine’s point is clear enough: “Sicque et latrones ad viam redeunt rectitudinis, et Iudaeus fidem amplectitur salvatoris,” (Thus even thieves return to the path of virtue and a Jew embraced the faith of salvation). The importance of this story to the overall narrative is perhaps not perceived in close reading. It is only with Voyant that we are reminded of the preponderance of the term iudaeus in the story, an indicator of its importance in the text. What can this tell us about mentality? St Nicholas is alleged to have lived in the fourth century. Is the story about the Jew contemporary to him or merely a product of the thirteenth century? Regardless of its actual provenance, its importance in Varagine’s narrative reflects the anti-Jewish fervour that was present in Europe at this time, which dated to the First Crusade (1095-99). Voyant’s “cirrus” view is a quick way of perceiving this preoccupation as it existed, whether consciously or not, in the author’s mind at the time of writing. This can be stated, I believe, as a general rule: the words and themes that appear the most in a text clearly preoccupied the author the most, and thus offer a window onto his or her mental world. In the case of hagiography, as stated above, one must nonetheless be wary of the use of literary topoi, which may indicate, more than one’s own mentality, efforts to evoke the mentality of Christians in the distant past. One last note here. The term iudaeus, as it appears in the “cirrus” view as the second most used term, is in its nominative form. A look at the text itself reveals, unsurprisingly, that the term is indeed also used in its other declensional forms, although these are not, for obvious reasons, factored in to the total frequency of the term iudaeus. This is another limitation of using Voyant with Latin texts. The true frequency, and thus importance, of a term in a text may be underrepresented in the tool’s “cirrus” view. This can only be remedied by manually searching, using Voyant’s “terms” list, for all instances of words beginning with iuda-. It is one extra step, although one that has the potential to significantly alter the tool’s results, especially if a certain term is used a number of times in multiple of its declensional variants.

Perhaps it is wiser to take a large group of texts from a single author and place these into Voyant. This will provide a more complete picture of the author’s mentality. I have done this for Varagine, inputting into the tool 23 vitae from his work, the “cirrus” results of which appear below in Figure 2. I have removed the pre-programmed Latin stopwords Once again, dei appears the most times, and a look at the concordance confirms that this is indeed, mostly “servant” of god or other related terms. The other thing of interest is the preponderance of the term corpus (body). It appears a total of 43 times throughout these 23 vitae. As stated above, we must however account for all the word’s declensional variants. A quick search for corp- produces 72 hits. In this case, Voyant’s inability to deal with a highly-declined language like Latin, has produced a quite substantial skew in the results. It turns out the word “body” is used much more in the texts than is initially apparent. Surely we can learn something here about the centrality of the body to medieval spirituality. Indeed, just as the world was saved through a body, that of Christ, so was the body heavy with diverse spiritual associations.[15] It is unsurprising then for the body to figure so prominently in these saintly texts. Without Voyant, this information would not so quickly and clearly be brought to the surface. That said, it would now require close reading to understand how the body functions in each vita, and this is perhaps, a more interesting question.

Figure 2

I would like now to turn to some more of the limitations in using Voyant Tools on medieval hagiographical texts. First, as Rockwell and Sinclair rightly point out, using Voyant on a text or group of texts runs the risk of divorcing these texts from their original author.[16] This, of course, depends on one’s willingness to accept, contrary to postmodern thought, that reaching through to the “author” is indeed an attainable and worthy endeavour. If we are going to assume this, then this form of text analysis produces some problems to that end. Students must be constantly wary of “finding” patterns within and between texts that do not necessarily correspond to the author’s intentions. It is not enough, for example, simply to conclude that the preponderance of the term corpus in Varagine’s vitae points to the centrality of the body to his spirituality. This claim must be qualified, using evidence external to the tool. It would be crucial, for instance, to consider Varagine’s education and career. Being a member and teacher of the Dominican order, the archbishop of Genoa was no doubt quite familiar with the doctrines and schools of thought current to his order in the thirteenth century. It is important to constantly keep such things in mind, rather than allow Voyant to “do the work for us.” This is also where we must remember Julia M. H. Smith’s emphasis on the interactions between author and audience. For whom was Varagine compiling his Legenda aurea? To what end? These are the types of questions that must accompany any use of hagiography with Voyant Tools.

Lastly, there is an issue inherent in digital text-analysis that must be addressed. To be frank, it is as Rockwell and Sinclair have put it, the fact that “computers see a text not as a book with binding, pages, ink marks, and coffee stains, but as a sequence of characters. The electronic text is a formalization of an idea about what the text is.”[17] In other words, Voyant reads and interprets what Voyant reads and interprets, nothing more. Things like marginalia can be neglected depending on the version of transcription. There is also (to my current knowledge) no way of digitally analyzing the parts of a literary work that are not explicitly textual. Indeed, Voyant suffers from an acute “logocentrism.” This is particularly problematic in the case of medieval hagiography. I am thinking here of Cynthia Hahn’s 2010 article in the journal Numen, “What Do Reliquaries Do for Relics?” Hahn is able to demonstrate for medieval reliquaries that, rather than serve to simply honour the saint whose corporeal remains are contained within, these artistic creations play an active role in “constructing the saint and his or her spiritual meaning for (and by) the viewer.”[18] If we apply this same concept to medieval hagiography, then suddenly the beautifully-decorated initials, as well as the, at times, countless illustrations that populate the margins of the text itself, become an important component in our understanding of the work as a whole. These are all things that Voyant cannot help with. Sure, the word “body” may appear countless times throughout the pages of the Legenda aurea. But how does the “body” figure in the parts of the document that are non-textual? Illustrations and visual aesthetics must not be forgotten amidst the countless hours one could spend playing around with Voyant and its attention to word patterns within and between texts.

It is also wise to keep orality constantly in the back of one’s mind when dealing with medieval hagiography. It was often the case that these texts were read out loud, especially on a saint’s feast day. In a predominantly illiterate society, this is in fact the medium through which a work like the Legenda aurea would be experienced in a majority of cases. Future work with Voyant and medieval hagiography must take this into account. A text becomes something very different when it is brought to life before an audience, such that patterns picked up through Voyant may be imperceptible to those listening.

In conclusion, Voyant Tools offers some opportunities to analyze mentality as it appears in medieval Latin hagiography. It is possible to glimpse recurring themes and thus to infer from them a correlation of some sort, whether positive or negative, with mentality. This is, however, not even half the battle. Scholars must account for the idiosyncrasies of the Latin language that sometimes significantly alter Voyant’s results. We must also be sure to include in our discussions reference to the author’s background, as well as the non-textual elements of the work. These considerations will help to ensure we are not getting lost in this new world of digital tool-assisted hermeneutics. Our medieval friends were, after all, writing well over five hundred years prior to the advent of the digital humanities.

[1] Felice Lifshitz has forcefully argued against the conception of “hagiography” as a separate literary genre. The term and its associated theoretical baggage are, indeed, merely modern positivist constructions, that mask the role of saintly literature as veritable medieval histories; Felice Lifshitz, “Beyond Positivism and Genre: “Hagiographical” Texts as Historical Narrative,” Viator 25 (1994): 95-114.

[2] For the centrality of the cult of saints to medieval Christian piety, see Peter Brown, The Cult of Saints: Its Rise and Function in Latin Christianity (Chicago: University of Chicago Press, 1981).

[3] For the diverse social functions of the cult of saints, see Simon Ditchfield, “Thinking with Saints: Sanctity and Society in the Early Modern World,” Critical Inquiry 35.3 (2009): 552-584.

[4] For the complex interactions between sacred biographical texts and their audiences, see Julia M. H. Smith, “Oral and Written: Saints, Miracles, and Relics in Brittany, c. 850-1250,” Speculum 65.2 (1990): 309-343.

[5] Manfred Gorlach, “Middle English legends, 1220-1530” in Guy Philippart, ed. Hagiographie. Histoire internationale de la litterature hagiographique (Turnhout: Prephols, 1994), 432.

[6] Andre Vauchez, Sainthood in the Later Middle Ages (Cambridge: Cambridge University Press, 2008), 3-4.

[7] Ibid., 535.

[8] Ibid., 536.

[9] See above, note 4.

[10] Eleni Tounta, “Saints, rulers and communities in Southern Italy: the Vitae of the Italo-Greek saints (tenth to eleventh centuries) and their audiences,” Journal of Medieval History 42.4 (2015): 435-37.

[11] Perhaps one of the most famous expressions of this school of thought is Peter Brown, “The Rise and Function of the Holy Man in Late Antiquity,” The Journal of Roman Studies 61 (1971): 80-101.

[12] Rachel Koopmans, Wonderful to Relate: Miracle Stories and Miracle Collecting in High Medieval England (Philadelphia: University of Pennsylvania Press, 2011), 26.

[13] Geoffrey Rockwell and Stefan Sinclair, Hermeneutica: Computer-Assisted Interpretation in the Humanities (MIT Press, 2016), 152.

[14] The University of British Columbia online Latin Dictionary,

[15] For the concept of Christian “materiality” see Caroline Walker Bynum, Christian materiality: an essay on religion in later medieval Europe (New York: Zone Books, 2011).

[16] Rockwell and Sinclair, 20.

[17] Ibid., 156.

[18] Cynthia Hahn, “What Do Reliquaries Do for Relics?” Numen 57 (2010): 311.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s