My impression is that some US universities at one time required producing a translation of a non-English language article in order to receive a PhD. As far as I am aware, this practice started before World War II when English was not as dominant a language in science as it is today. The requirement seems to have been removed at some point after World War II, and at some universities not until after the Cold War ended.

Does anyone know about the history of this requirement? Why was it added? Why was it removed?

Not much can be found online about these sorts of requirements. UC Berkeley has a page on what sounds like a similar requirement, but only for some programs. And it sounds like the concern is more knowing a foreign language than necessarily producing a translation. Perhaps producing a translation was only one way to satisfy a (now removed) foreign language requirement.

I recall speaking with someone who got a PhD in math from NYU in the 90s. They were required to write a translation, though as I recall they described the process as more of a formality and did not believe it continued long beyond their time at NYU.

Also, I spoke with a European academic interested in translations who was not aware of similar requirements in Europe. If anyone knows of similar requirements in Europe and their history, I'm also interested in this.

  • 4
    @Sverre I'm not surprised that a linguistics department requires knowledge of foreign languages. The OP notes math specifically. There it's less obvious. Jul 6, 2021 at 15:38
  • 20
    Before the early 1990s, virtually all U.S. mathematics departments with a Ph.D. program in (pure) mathematics required TWO foreign languages for (pure) mathematics. In fact, of the roughly 20 such departments I was initially looking at in Spring 1989, only two required just one foreign language, one of which I wound up attending. In my previous graduate attendances (yes, there was more than one; four in fact) earlier in the 1980s, I don't recall ever seeing a program that required just one foreign language. The languages usually had to be chosen from French, German, or Russian, (continued) Jul 6, 2021 at 15:52
  • 4
    but others could be used with permission. For example, Italian was a common alternate language asked for. What it took to satisfy the language requirement varied greatly, however, and sometimes even within the same department (because often one could propose a way of satisfying one of the language requirements, which would then require departmental and/or Ph.D. supervisor permission). Sometimes translate a paper or two over a weekend, sometimes translate a paper and/or passages with use of dictionary alone in a somewhat supervised empty room, etc. (continued) Jul 6, 2021 at 16:00
  • 10
    Note that historically in much of Europe, this was the case for any higher education. Latin and Greek were essentially mandatory to be considered a well studied scholar, and if you were a scientist you probably also knew some combination of German, French, Italian, and English. This wasn’t some arbitrary decision, but a result of the relevant texts being in those languages and people being sensible enough to consider it better to read the original than a translation. Jul 6, 2021 at 23:23
  • 12
    As late as the 1970s, Cambridge University (UK) required all undergraduate students (in any subject) to have passed examinations in a modern foreign language and also Latin. Of course having a national high-school-level examination system in the UK meant they didn't have to test applicants themselves. Probably Oxford had the same requirement, but I'm not sure. But this was no big deal, because anyone attending a UK Grammar School and aspiring to go to university would have taken the two exams at age 16.
    – alephzero
    Jul 7, 2021 at 2:59

7 Answers 7


Some personal evidence.

In the 1960s I had to show that I could read mathematics in at least one of French, German or (I think) Russian. That was a reasonable requirement at the time - much historical literature that was still relevant called for that kind of literacy.

The exam did not call for a formal translation. I just had to make reasonable sense of a page or two of mathematics - not highly specialized. Presumably if I could do that I could master the details of any paper I really needed to understand.

  • 4
    Can you really be older than I am??? Anyway, my experience, a bit later, was both French and German. There were courses especially for grad students. The "proof" was a sufficient grade in the class or, perhaps some other demonstration for those who were already literate in one or more of these. Likely a brief oral exam. But it wasn't long before the need for that became moot as the academic world moved more toward English. IIRC, Russian might have been a substitute for one of the languages and later became one of the preferred choices.
    – Buffy
    Jul 6, 2021 at 14:26
  • 39
    @Buffy I'm 83. So yes. Jul 6, 2021 at 14:29
  • 5
    You are having a good run, as am I. NO surrender.
    – Buffy
    Jul 6, 2021 at 14:35
  • 9
    This agrees with my experience, even though I'm only 73. Jul 6, 2021 at 14:41
  • 3
    I’m “only” 64 and for my Ph.D. (U of Wisconsin) had to produce a written translation of several pages of mathematics from two of French, German, and Russian. Each test was a few hours long, and the use of a [language]-English dictionary was allowed. If memory serves, one could petition for another language if there was significant literature in that language for the student’s specialization. Even now, on occasion I find something worthwhile that’s only available in the original French or German and can slog through. I suspect Google translate isn’t ideal for mathematics but haven’t tried it.
    – Steve Kass
    Jul 6, 2021 at 23:50

To comment on the "why": at a certain point a lot of work in many fields, including mathematics, was untranslated from the original into other languages. There was lots of critically important mathematics written in French and German, among other languages.

Russia did important work also, but a lot of it was hidden from much of the rest of the world due to the Cold War. Some things actually had to be independently replicated in Russia because of the Cold War. St. Petersburg (Petrograd, Leningrad) was one of the centers of deep math thinking. IIRC, the Simplex Method was discovered independently there.

The same was true of other fields with a bias toward those same languages - philosophy, for example. But Spanish or Portuguese, maybe Italian, might have been more critical for someone studying some aspects of Literature, though translations of non-technical work came a bit earlier than scientific work.

But translation is more ubiquitous now and there is a bias toward English publication that didn't exist before. If you are a scholar in Germany at the moment, you are likely fluent in English. Maybe more so that lots of native speakers.

  • There still is a lot of important mathematics in French, which hasn't been translated, and I had to read a German paper for my research last month (although that was admittedly a special case) Jul 8, 2021 at 12:51
  • I am looking for a reference for a fact that I think must be in the literature, and I have just printed out a paper by Demazure (in French of course!) which is one of the likely candidates to contain it. Jul 10, 2021 at 22:30

A few years ago I pushed rather hard, with ultimate success, to end one of these language requirements at an R1 university math dept as a grad student representative. More than anecdotes would make for a better answer, but I've never found an authoritative synthesis on the history of the requirement or even a careful summary of its current status. Perhaps someone will find what I managed to gather useful.

My recollections are from 2018, exclusively for Math PhD programs. They're nowhere near authoritative, but I gave soliciting information from a wide range of people the old college try at the time.

Why was it added?

I've never seen a historical source for the original motives, but talking around both locally and more broadly at conferences, I never heard any deviation from what other answers report: Translation from other languages proved academically useful in the past because a significant proportion of relevant material was not written in English.

What is the status of this requirement?

The status was all over the place in 2018. For some hard names, here's a list of schools I compiled at the time for political purposes, though perhaps they've changed now.


  • University of Illinois Urbana-Champaign
  • Georgia Tech
  • UC Santa Barbara
  • UC Berkeley
  • Florida State University
  • Duke University


  • UNC Chapel Hill
  • University of Miami
  • University of Virginia
  • Harvard
  • Princeton

What I gathered from local observations was that the requirement has been on a slow rather than abrupt decline. Older academics occasionally reported specific examinations were more extensive than modern ones, e.g. requiring two languages rather than one. The history of the specifics differs on a per-department basis, of course.

The requirement has certainly not been entirely phased out everywhere at time of writing. For instance, despite my success at ending the requirement in my Dept in 2018, that was only for future PhD classes- I and my peers still had to take language exams. Students with language requirements are still on the books.

Why was it removed?

Writ large, the argument that worked was the obvious: The proportion of relevant material not written in English has declined to the present day.

Some secondary points that arose:

  • Requirements incur a cost, namely paperwork and occasional graduation delays.
  • The specific languages (usually French, German, and Russian in math) were vestigial remnants of the original intent. While there was some suggestion that we could debate a change to the list, ultimately this would have induced more work.

Why has it lasted so long?

On the one hand, what I gathered from my investigations wasn't revelatory in this respect. Bureaucracy moves slowly. Democratic mechanisms like faculty votes demand energy to change the status quo, whatever it is.

An interesting secondary point I gathered. Whether it generalizes to other Universities, I can't know for certain, but I have the impression that most operate in a structurally similar way. In Departments where the work to maintain requirements mainly gets placed with short-term faculty directors (of graduate studies, etc.), almost nobody has an immediate incentive to remove outdated degree requirements. The most burdened voter, the director in question, will probably have to deal with the removed req for another 5+ years. That's longer than most graduate directors' terms.

  • 2
    Yes, indeed, the (dis) incentives for change are significant. Some (older!) faculty have claimed that "it'd be good for students to ...", and my rebuttal was that, while it might be good to have early-morning calisthenics for grad students and faculty each day, I don't think we should require it. :) Jul 6, 2021 at 22:14
  • 1
    "...the (dis) incentives for change are significant." Or, simply: "...all experience hath shewn, that mankind are more disposed to suffer, while evils are sufferable, than to right themselves by abolishing the forms to which they are accustomed." Sure, from a slightly different context, but otherwise seem applicable. :) Jul 7, 2021 at 5:37
  • 2
    "Why has it lasted so long?" You mean aside from the obvious, intrinsic value of having your students/graduates fluent in multiple languages that are historically popular among academic literature authors? There's probably also the perceived benefit to some (e.g. Harvard, et al) that it separates academics into an 'elite' category vs the common rabble.
    – TylerH
    Jul 7, 2021 at 13:48
  • 1
    Things are different with each decade of mathematical knowledge advancement (and specialization) and, recently, availablity of automated translation. But I feel I was a better mathematician (I am one no longer) by having made an attempt to read Bourbaki in the original, ditto M. Gromov, and some Germans I forget. It is a different form of discourse and reasoning. It was easier for me being multilingual, for sure, and not be important enough to be a Ph.D. requirement. But a math professional should at some point become aware of the history of their field, and this can be an element of that.
    – Houska
    Jul 7, 2021 at 19:55
  • To @TylerH 's comment: I've tried not to overload this answer with a blow-by-blow retelling of the political side of my experience on the topic. To contextualize slightly: Your point and many others arose transiently. But I also observed a general sentiment of surprise that the req was still around and think highlighting that and its potential causes more pertinent to the question. Are you suggesting a change to the answer?
    – user137975
    Jul 8, 2021 at 0:46

Echoing other answers, with a few more details:

Yes, in my own experience and direct observation, it was typical at U.S. R1 universities, in mathematics, to require "reading knowledge" in two of French, German, or Russian. As others have said, this was because there simply were no English-language primary sources (nor translations) for lots of important things. I think it's fair to say that German and French were more important languages for mathematics (not counting Latin!) from 1770 until 1945 or so.

For example, in the mid 1970s, most of J.-P. Serre's high-level expositions from College de France and elsewhere were only available in French. Some of them started being translated into English in the late 1970s.

Much of C.L. Siegel's work was written in German, and was not translated. Indeed, in those days, so many math people could read German that there seemed to be no reason to translate it.

(In contrast, relatively few people in the U.S. and western Europe could read Russian, which surely helped motivate the Amer. Math. Soc.'s translation program for various Russian sources.)

In 1977-79, P. Deligne's contribution to the Corvallis Conference (held in Corvallis Oregon) (AMS Proc Symp Pure Math 33), was written up in French. At the time, this seemed a little idiosyncratic for an English-language conference, all whose other contributions were written up in English. But no one got excited about it, because essentially everyone in the audience (at the conference, and for the conference proceedings) could read/understand some French, at that time.

In recent years/decades, the utility of being able to read more than in-English mathematics has decreased, not only because more things are written in English, but, also, because translation software is often more efficient than a weak grasp of the languages (moderated by some sanity checking about technical things).

So, at my R1 U.S. univ, in math, we had reduced the language requirement from two to one some years ago, and will reduce it to $0$ in the immediate future.

Yes, this does cause some bottlenecks for my PhD students, because it's still not the case that everything is available in English. I am indeed quite glad that I can read math in French fairly easily, and in German with some effort. I regret not learning a little Russian when I was younger, too, but it was rarely available in the U.S. in those times.


I did my Ph.D. in ~2000 at a math department (in the US) that required demonstrating reading proficiency in another language, as evidenced by being able to translate a mathemetical article in your field or a closely related one, to one of the professors in the department who was professionally comfortable in that language.

There were those who preferred to do so by writing out a translation of their chosen article. Others preferred to show up with the article and viva voce translate at the random spot chosen by the faculty member, say 1/2 page worth.

Our general departmental culture was very collegial, so if you did at all a creditable attempt, you passed. Just like it was hard to actually fail our comprehensives. There were ongoing friendly arguments whether the oral or written approach was "better", which depending on the disputant could mean "easier" or "more creditably demonstrating sufficient mastery".

The arguments in favour of "written", which seems to be the context of the question, were from both angles:

  1. An acceptable written translation better demonstrates you sufficiently understand nuance, and provides documentary proof thereof (can be filed away).

  2. (From the other side) Written can actually be easier, since you can look up in a dictionary, take your time, redo part, as opposed to "being on the spot" orally.


You asked for a contemporary example in Europe: Russia still requires PhD theses to be translated into Russian in order to get foreign PhDs officially recognized in Russia.

Apart from that, there is a list of foreign universities whose PhDs get automatically recognized in Russia without translation. There is also list of Russian universities that can recognize foreign PhDs on their own, without approval of the central authorities (VAK). However, neither list is particularly long, and not all universities that have the right to recognize foreign PhDs on their own also make use of this right.

  • 2
    I think the question is about requiring the PhD candidate to translate a foreign research paper into their own language, but your answer is about requiring the PhD candidate to translate their own thesis?
    – Stef
    Jul 7, 2021 at 13:21
  • Yes, indeed, it seems that I misunderstood the question. My answer is about translating the thesis itself.
    – user33593
    Jul 7, 2021 at 16:31

An anecdote from a relatively recent experience:

I completed my PhD (pure mathematics) in 2017 at a Big 10 university in the Midwestern US. At that time I was required to complete a language exam in either French, German, or Russian, none of which I knew anything about. I decided to go with French since I do know some very basic Spanish, and the syntax is fairly similar.

The graduate chair told me I was to find a 10-page excerpt of an appropriate mathematical paper or textbook and take it to a specified French-speaking mathematician in our department for approval. I selected a chapter from a text on the differential geometry of surfaces, and it was approved.

The French-speaking mathematician then told me to take a few days and do my best to create an English translation of the excerpt. I was permitted to use whatever resources I wanted, even translation software. I did it mostly with Google translate and the assistance of a standard English-to-French dictionary.

I took the original excerpt and my translation back to him, and we reviewed them together. He asked me a handful of questions about my translation, and also asked me to read specific passages from the French original and describe what they said without referencing my translation. The whole meeting was very informal and took less than 10 minutes.

The point was basically to demonstrate that you have the ability to understand mathematical work not written in your own language, and I did that, so I passed.

This requirement was mildly annoying but not very burdensome, so students in our department have generally not pushed back much against it. Also, I believe (but I'm not completely sure) that students who speak English as a second language are exempt.

A final comment: I know of a pair of PhD students at another Midwest Big 10 university who petitioned their department to allow them to take their language tests in Italian. They were studying PDEs, and apparently much of the relevant literature to their work was written in Italian, rather than French, German, or Russian.

  • I have also heard that some algebraic geometry PhD students took language tests in Italian, because the work of the Italian school of algebraic geometers was relevant to them. Jul 8, 2021 at 17:31

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .