Humanities + STEM: Ethical Consumption before Capitalism | An Interview with Dr. Astrid Giugni
Dr. Astrid Giugni (English and Information Science + Studies) leads the Bass Connections project “Ethical Consumption before Capitalism,” which brings undergraduate STEM and Humanities students together to trace how the discourse of commercial consumption and the labor needed to sustain early modern markets are presented in the pamphlets, sermons, and meeting records produced for and responding to the English Trading Companies. See https://bassconnections.duke.edu/project/ethical-consumption-capitalism-2024-2025/
Joshua Palomo, a double major in Medieval and Renaissance Studies and Statistical Science (class of 2025), interviewed Dr. Giugni in depth about her exciting research project in the field of Medieval and Renaissance Studies.
Do you mind describing what the goals are for your Bass Connections project “Ethical Consumption before Capitalism,” the time period you’re exploring, and what you’re trying to investigate?
This Bass Connections project is heading into its third year, but it’s been going on as a Data+ summer project for five years. The idea of the project is to narrow down on texts that have to do with the Virginia Company of London, which was a monopoly in the 16th to early 17th century directed at the economic exploitation and colonization of the New World. We are looking at how the Virginia project is discussed in literature promoting and describing the New World—so sermons, descriptions of the American Southeast, the people who are indigenous to this area, the climate, the commercial opportunities. We’re also looking at the Bermuda Company, an offshoot of the Virginia Company—as some of the key figures from the Virginia Company eventually become deeply invested in the Bermuda venture.
Our aim is to understand the time period better and how the push toward Virginia was being described and advertised. The company was not financially profitable, making it harder to justify its continued existence as a commercial entity with a monopoly. We’ve been tracking how the Virginia Company’s language in its public documents evolved over time from a commercial focus, promising that the Virginia venture would be highly profitable, to an explicitly religious emphasis on countering Spanish influence and Catholic proselytizing in the New World. As economic profits fail to materialize, the documents we are studying develop a language of accruing a “profit of souls” (referring to the potential conversion of indigenous people) by turning them to the correct form of Christianity, which for English settlers meant Protestantism.
At the same time, settling Virginia is seen as a potential solution to the problem of England’s population growth. In particular, at the turn of the 17th century, London can’t keep up with providing housing for its burgeoning population. There is an exploitation of tenements and overcrowding of available housing, but also there is not enough employment. One idea that starts being pushed is that Virginia could be promoted as a sort of escape valve for this demographic pressure. The New World could provide abundant “empty land,” so much so that by 1618 the City of London agrees to pay for the transportation of “vagrant children” from the city streets. The City engages in a short-lived but agressive program of finding, arresting, and confining children and young adults to places like Bridewell before sending them overseas. The City government would pay £5 per “vagrant” youth to be sent to the New World to work as indentured servants, with the promise that at the end of their indenture they would obtain a certain tract of land.
This is the general picture that motivates our project—to better understand the motivations and consequences behind the changing explanations for the Virginia Company’s self-advertising, requests for funding, and support from the Crown as well as from the people of London and their investments.
I want to get a sense of how you gather your data. Projects that I’ve worked on tend to use a specific database like EEBO (Early English Books Online), but I want to know how you collect your information and how you deal with inconsistencies or incompleteness in your datasets.
Part of the challenge is always trying to patch in gaps in a given database. We have a very heterogeneous set of data we’ve been trying to work with. First are printed works advertising the Virginia Company that are found through the EEBO Text Creation Partnership. This gives us access to manually transcribed, digitized works that are quite useful for large-scale natural language analysis. For example, we’re tracking language found in a series of sermons that the Virginia Company commissioned as a form of advertising and fundraising. This advertising language can be traced from the 1609 second Virginia charter issued by King James I through to the 1622 reports of the attack on Jamestown, when hostilities were reignited between the indigenous people in the Powhatan confederation and the settlers.
That’s one source of information. A second one includes the records of the Virginia Company, much of which are available in digitized form online. We can see the minutes of each meeting, which provide a roster of who was present that day, who wasn’t, what topic they were discussing, which funds they were using, etc. These are the more computationally tractable sources of data.
Last year (2023-24), however, we tried to work with online manuscript sources including parish records to obtain information about the “vagrant children” sent to Virginia as indentured servants. We tried to match the parish records to some of the Virginia Company records and to a printed edition of the Bridewell records for our period. For instance, if we have a Bridewell record with first name, last name, and some other information—sometimes the entry will say “children of,” “from some village in Somerset,” whatever—we can try to find the correct church that would have their date of birth and other information. The hope was to find out more details about the children forced to emigrate: how old they were or if they had other family in the New World. This was the most challenging part of last year’s work.
We try to fill in the gaps as much as possible. The Bridewell Hospital records sometimes just contain lists saying, essentially, “constable so-and-so picked up 20 children during this week in this area,” so there was nothing further we could do without names. You can actually see that, at the beginning, these records are much more careful about tracking the names of children, but then later on they simply list the number of “boys” or “wenches” arrested. We were able to locate some of these child “vagrants” geographically in the area of London where they were arrested, but that’s all the information we had about them. One thing we’ve been concerned about is how much detail was being tracked at the time—how much care was given to keeping track of these children versus “we just need this number of workers in Virginia.”
Tell me more about how you analyze this data. What does the general research process look like for students working on your project?
We track specific terms; using models that turn each word into a vector, you can then figure out how close two vectors are. This, in turn, gives you a sense of the meaning of a given term in its context. So we can see, for instance, if the word “soul” is being used to mean the spirit of a person: whether its vector is closer to the vectors of other spiritual language like “Christ,” “God,” “heaven,” and “salvation” or to the vectors of words like “farmer” or “payment,” pointing toward economic language such as “20 souls to labor/farm.” This allows us to track at scale and over time whether the documents associated with the Virginia Company focus more on spiritual language or sometimes on economic, farming, or labor language. Because the Virginia Company, a joint-stock corporation, was not a monolith, we can see how language use differs among authors—say, that of John Donne, who preached a sermon before the Virginia Company but never visited the New World, versus somebody like Patrick Copland, another minister who gave a sermon for the Virginia Company, who raised funds for schools in Virginia and eventually settled in Bermuda.
The details of our analytical approach really depend on students’ experience both with the data science and the historical aspects of the project. Some of the students—especially those who have done Data+ with me as well as Bass Connections and have been engaged in the program for months—I let them roam free because they understand what the goals are, what the techniques are, and I know they can work independently. I give newer students specific tasks, for example, organizing some data in the records of the Virginia Company we’re currently working on. At the moment, we are trying to organize the London meetings chronologically for all of the dates that we have and present them visually, but it’s very hard to see patterns if you go day by day, and going year by year creates buckets that are too broad to put the data into. After a group discussion, we figured out how we could clump the data according to seasons, but not by the calendar seasons of fall, winter, spring, and summer, but rather according to sailing and agricultural seasons, because that’s what would matter to the colonists—Are you going to arrive in the colony during harvest versus during winter when there is nothing to eat?
Depending on students’ interests and skills, I try to match them up to a specific task at the beginning or during crunch time of a semester when we’re busy, but ideally, if they have a particular interest, I let them run with it. The students who participate in these projects are here because they are interested in learning how to conduct interdisciplinary research, and I welcome their independent engagement with this work. Over the summer with Data+, one of the students got interested in matching phrases or words to specific authors and creating a network analysis illustrating how certain factions in the company shared linguistic habits. It was the student’s idea and I and Dr. Jessica Hines (the co-leader of the Data+ team) said go for it, because that sounds like a great idea and we would love to see the result.
Student creativity is key to the project. While the project definitely has an overarching idea and specific goals, when working with historical data you run into gaps and problems, and students’ input is crucial to finding good solutions. For instance, we were hoping to use GPT’s API to help us clean some text we OCR’ed [Optical Character Recognition], but it kept crashing on us, so we had to switch gears and do a text analysis on the “dirty” data, meaning we did not clean out artifacts from the OCR software. We have the code designed, and we have a process and workflow taking our documents from images to OCR’ed text so as to be able to perform larger-scale analysis. During the summer with Data+ we can pick it up from there.
Do you work with students who have different backgrounds in terms of level of comfort with coding or historical analysis? Do they seem to share an interest in the time period or a preference for the coding aspect of the project?
A mix and all of the above. I see a lot of students interested in computer science or data analysis who want to see NLP [Natural Language Processing] scenarios, because in classes you often get examples, but that’s very different from real life with missing data, etc. But I also see students whose primary interest is in the historical period. I try to match student interests as much as possible, because I think it’s helpful to involve students who have experience working in premodern periods as well as students who may be strong with Python or R. The combination of interests is valuable, especially once you get to the visualization stage of a project; if you have tried different types of visualizations, you will feel comfortable experimenting.
Has your team presented this research anywhere?
The recent group has not presented yet because we’re using a larger dataset than we did for the very first project I did with Bass Connections. One earlier group that looked at late-17th-century political economic texts—that was one of the starting points for this project—ended up presenting their research at the Northeastern Modern Language Association Conference in 2021. From what I heard from the students, it was an exciting experience because it was a poster session where they got to have both faculty and other students walk up and ask them about their project, and they ended up winning a prize for their presentation. They were proud of what they did understandably. I was delighted because they did really great work.
What’s next for the current project?
We’re going to continue this summer with Data+, cleaning and expanding the data that the students prepared in the spring term about geographical locations in the records and people involved in specific sub-ventures of the Virginia Company. We will build on the pipeline designed by the Bass Connections team to further clean and analyze this large dataset, and then find a good way to present this visually in geographical and temporal terms. Essentially, we have a lot of data that is only partially analyzed—we just need to figure out a way to finish the analysis and present it cleanly and clearly.
Eventually, we would like to apply for external financial support for the project to allow us to extend our analysis and fund a larger group of students. I believe students should be supported financially for project work like this as much as possible, and I want to make sure they can travel to archives to do research and to conferences to present their findings. I’m taking my group this year to the University of Virgina to look at their land records and manuscripts that are not available to consult online.
Is there anything else you’d like to tell me about what you’ve enjoyed about this program?
I have tremendously enjoyed seeing students stretch their expertise from the humanities to the computational side or from the computational side to the humanities. Having a full year to focus on a group research project allows students to really build on learning done in the classroom. You might be great at statistics and be a great programmer, but getting to apply that with students who are humanists is spectacular, and the results are always great.