After a long run of somewhat pedantic articles on my favorite sub-sub-specialty, I thought I’d take an opportunity to indulge myself a bit with a different topic. I’d like to make a case for why, in an age of near-instant and reasonably accurate translations provided by the likes of ChatGPT and Grok, it still behooves the student of the Middle Ages to learn languages, including modern ones [1].
But first, a personal detour. I have a vivid memory of sitting in the Medieval Institute, roughly 12 years ago, and making a Discovery. The subject of my dissertation, Humbert of Silva Candida, had not directly quoted Epiphanius of Salamis, as implied by his text, but rather a redaction of the De haeresibus of St. Augustine now known as the Liber Praedestinatus (possibly the work of Arnobius the Younger). This attribution had escaped Cornelius Will, the most recent person to have published an edition of Humbert’s works on the conflict between Rome and Constantinople, and I was exceedingly pleased with myself. I just as clearly remember, a few months later, sitting in the same room and discovering an article written by Anton Michel, a well-known scholar in my field who was active in the first half of the 20th century, in which he made exactly the same identification a century, to the year, before I had [2].
Over the course of the same project, I made a second, and even better, discovery. One of Humbert’s works quotes a passage in Greek and attributes it to St. John Damascene. The text is hopelessly garbled in the published edition, but a few of the manuscripts preserve the Greek a little bit better and some even have an interlinear gloss translating it into mostly comprehensible Latin. After an inordinate amount of time digging around, I was finally able to identify the source – one of the variable texts used in the Orthros (Matins) service of the Byzantine rite. This was, in the words of one of my dissertation committee members, a “real discovery”, and I was sufficiently impressed with my own sleuthing that I incorporated this work into an article a couple years later. Of course, the nature of the academic world is that mistakes you don’t catch yourself are often caught for you, and the corrections are printed, public, and permanent. In my case, a scholar at the University of Bologna pointed out that the same identification had already been made by Georg Hofmann way back in 1950 [3].

Upon some reflection, I think that the common theme in both of these misses—one caught in time, the other not—was a failure to grapple sufficiently with secondary literature written in German, which includes the work of both of the scholars mentioned above. This failure, in turn, was brought about by my tendency to avoid German sources, stemming from my less than fully confident reading abilities in that language. Seeing a book written in Fraktur still sends me into conniptions. I was uncomfortable with the language, and so I avoided anything that did not have an obvious bearing on my own research. But, as it happens, it is not possible to ignore German scholarship, not even a little bit, without being embarrassed. Hence, my plea to current and prospective students in the field: do as I say, not as I have done.
Learn your languages properly. And this is where the formal requirements of a graduate program (for example, those of Notre Dame’s Medieval Institute) are necessary but insufficient. Typically, proving reading language proficiency for academic purposes requires being able to complete a close translation of a prose selection in the target language—a page from an article in French, for example—with the help of a dictionary. This is precisely the level of language ability required to be able to quote a passage from foreign-language scholarship or to double-check a translation rendered by a large language model (for the less pure-minded), which are indeed absolutely necessary skills. They are, however, a far cry from a comfortable working knowledge of the language in question. Instead, I’d propose what I’d call a “library stacks test” as the first part of an informal guide: can someone pull a book in the target language that’s related to their field of study off a shelf, spend a minute flipping through it, and have a reasonably good idea of the contents and how it relates to his or her work? This then would be followed by a reading test – is it possible to read the book (or, as we used to say while preparing for our comprehensive exams “to familiarize oneself with a book”) and be able to incorporate it into a paper in a reasonable amount of time, that is, in not too much longer than it would take for a book of similar length written in English?
These are both tests in which, fortunately or not, the abilities of large language models (for those inclined to use them) fail to deliver much assistance. Microsoft CoPilot, even when not suffering from a global outage, is unable to give an opinion in the moment on the utility of something written in Italian that you happen to discover right next to the book you were originally looking for on the shelf. To rely on LLMs, then, is to abandon the potential for so many serendipitous discoveries made in the library stacks. And even if we allow that an LLM can deliver an adequate translation of a book, there remains a mountain of secondary literature in the field that remains undigitized and thereby untouchable without a serious effort to digitize and organize it. In practice, I would suggest that this is a significant enough barrier to prevent ease of access, which is the core of the problem.
So what would I recommend to a prospective student (bearing in mind that I didn’t do this myself)? The Medieval Institute at Notre Dame requires one medieval research language (typically Latin) and two modern research languages (often French and German). Latin (or Greek, Arabic, Hebrew, etc.) is a topic unto itself and deserves separate consideration. For the modern languages, though, French and German do probably cover a fair bit of the secondary literature for most of the disciplines represented in an interdisciplinary program for medievalists. So far, so good. As I suggested before, though, it is the level of familiarity and comfort that is at issue. For that, I’ll go out on a limb and suggest that a B2 level, or upper-intermediate, using the Common European Framework of Reference for Languages, is probably about what we ought to aim for. I’ll go out on another limb entirely to suggest that this include spoken proficiency, as well as reading. There are very many superb scholars who don’t speak a lick of anything other than English, but it’s comfort we’re going for here, and I think that spending a couple hundred hours chatting with tutors, friends, or strangers in the target language is the most efficient way to get there. By way of anecdotal evidence, I know of several other students, wiser than I, who took the opportunity to acquire a spoken command of their research languages. I’d be willing to bet any amount of money that none of them regretted it [4].
While this appears, admittedly, a formidable hurdle for many American students, reared, as I was, as monoglots, it is by no means insurmountable. The Foreign Service Institute, run by the US Department of State, suggests that it takes a native English speaker about 600 hours to learn French and 900 hours to learn German to a level of professional proficiency. I can’t speak for German, but my experience with other Romance languages outside of my academic study suggests that it might take a little less time than that, especially for a motivated student focusing deliberately (and with a background in) specialized academic vocabulary. But all that is to say, it’s a manageable goal, particularly for the student who has the good fortune to start in college or high school (or earlier). It’s also never been easier. We live in an age of true abundance when it comes to language learning materials: a plethora of apps (admittedly, of widely varying quality), online tutors, and a legion of YouTubers making “comprehensible input” style videos. Formal classroom instruction and immersion trips can be nice, but they are by no means necessary [5].
I’m late to the game myself but now, a decade overdue, trying to make up for lost time. I can only recommend that you, O reader, start sooner.
Nick Kamas
PhD in Medieval Studies
University of Notre Dame
1. I should be clear about my target audience: I mean to address very junior or aspiring anglophone scholars. I wouldn’t dare presume to lecture anyone else.
2. Anton Michel, “Praedestinatus, eien ungenannte Quelle Kardinal Humberts im Kampfe gegen Kerullarios (1053/1054)”, in Festgabe Alois Knöpfler zur Vollendung des 70. Lebensjahres (Freiburg im Breisgau: Herdersche Verlagshandlung, 1917), 240–247.
3. Along with noting a rather embarrassing transcription error that I had made. Luigi D’Amelia, “‘Translating Bread:’ Notions of Etymology and Theory of Meaning in the Latin-Greek Controversy on the Azymes”, in Latin Translations of Greek Texts from the 11th to the 13th Century (Leiden: Brill, 2025), 302.
4. I write this line with my old friend, neighbor, and early mentor Steve Molvarec in mind. May his memory be eternal.
5. And may or may not even be all that helpful, depending on the instruction methods employed. For this and the notion of “comprehensible input”, see the work of Stephen Krashen, mostly available here.







