Friday, February 7, 2014

Language and Math… and a PhD. (and Navier-Stokes)

Je l'aime quand vous parlez français....*

Perhaps I've just been too long outside the loop of higher academia, but here's something I was completely unaware of:
"...a little-known fact: to get a PhD in maths from Harvard, you need to be a language buff. The university points out that 'almost all important work' is published in French, German, English or Russian, and so 'every student is advised to acquire an ability to read mathematics in French, German and Russian'. This makes it sound optional, but if you want the PhD you have to pass a two-hour written exam in two of these three languages."
That's the beginning of a new piece in the New Statesman which is actually about the recent claim of a Kazakh mathematician to have proven the generality of Navier-Stokes equations, one of the Millennium Prize problems. Because of the complexity of the proof, and the fact that it is written in Russian, it will take significant time to confirm, and so we face what the author calls "a mathematical pile-up at the language barrier."

The article is interesting for its brief discussion of Navier-Stokes, but I was more struck by the "little known fact." Assuming the author is correct, and a knowledge of two languages outside English is indeed a requirement (not merely highly-advised) for the Harvard math PhD., I'm curious if this is now the norm at most top-flight mathematics graduate schools, or does it vary considerably from PhD. program to PhD. program? Also, does the "ability to read mathematics in French, German and Russian" perhaps entail significantly less proficiency than would be required for fluency or conversational ability in the languages?

(*  "I love it when you speak French" ;-)


Evelyn said...

I can only speak for my own experience, which is not Harvard. We had to pass an exam in one of French, German, or Russian. It was one hour, and you translated a mathematics paper in a related area to your area of study with a dictionary. I took Latin in high school and college, so I chose the French exam and had very little trouble with it even though my French is practically nonexistent.
At my school, it seemed like the language exam was becoming more and more a formality. The fact is that these days English is the language of mathematics, and the importance of reading French, German, or Russian is very low. I think some schools may be phasing this requirement out, but that is just a guess. In the past, I'm sure this was much more important.

My impression is that most math PhD programs require one language exam, often a translation like my school's, and some require two. I'd be surprised if Harvard's requires fluency in either language. Reading is much easier than speaking or writing, and familiarity with English makes both French and German, but especially French, very easy to read. (In French, many math words are the same as English with a different ending. German seems to throw more curveballs.)

"Shecky Riemann" said...

Thanks for the input Evelyn. It's difficult for me to imagine that the Harvard standard (as described) would apply across-the-board to other top schools.

Evelyn said...

After I left my comment, I read Harvard's degree requirements. It sounds like their language exam is similar to the one I took, although they have to do two languages instead of just one. They're asked to translate about page of mathematics, and they have access to a dictionary if needed and two hours instead of one (ours was more "what can you translate in an hour" rather than a goal of doing a certain length of translation).

I haven't asked any of my Harvard friends how strictly the exam was graded or how seriously it was taken, but it sounds like it was probably similar to ours and is probably similar to most schools'.

I also think the article is incorrect that the lack of mathematicians fluent in Russian is delaying review of Otelbayev's paper. There are tons of Russian mathematicians working in the US, and some of them study Navier-Stokes type stuff. From some rumors I've seen, it looks like there may be some insurmountable flaws in the paper.

"Shecky Riemann" said...

yes, the article seemed a little off-base on some things, including there being any shortage of Russian-speaking mathematicians (although Navier-Stokes is no doubt a pretty specialized area). Anyway, interesting to already hear of 'rumors' of 'insurmountable flaws,' though still early in the process.

E. Littlestar said...

Hello ,
I've studied mathematics and physics and I know English , French ,Russian and a little German.I could translate Otelbayev's paper for myself ,or for others .I could find time to translate it and some more time to understand it or get deeper into it ,although for now I have a general knowledge of the Navier-Stokes equations.In any case if I make any progress I'll make it known.