[Computational Complexity] On Language
Language has never been my strong suit. I didn't speak full sentences until I was five. I had a 220 point spread between my verbal and math SAT scores. I fumbled through three years of high school French (which required some summer school). This knowledge of French was only useful a couple of times. Wandering the streets of Paris, a women asked me Quelle heure est-il? and I knew enough to show her my watch but enough to actually tell her the time. Also I saw Secrets & Lies in France and sometimes the French subtitles made more sense than the heavily accented English.
During my undergraduate years at Cornell I struggled and gave up on Spanish. Luckily a linguistics professor had a theory that people who had trouble learning English early (like me) would have too much difficulty in picking up a new language, so I could take an intro linguistics course to cover my language requirement. Pretty cool as we covered context-free languages simultaneously in linguistics and in my introduction to theoretical computer science class.
In graduate school my three years of high school French got me out of the Ph.D. language requirement. If English was not the lingua franca of our field, I would be in serious trouble. I've always been impressed how many non-native speakers of English have succeeded in computer science.
I spent an entire year on sabbatical in Amsterdam but only learned enough Dutch to navigate the supermarkets and order in restaurants. Most Dutch speak English (and 3-4 other languages) and my attempts to say most Dutch words usually got responses in English. Still I definitely missed something as when I left a conversation the language shifted to Dutch and I couldn't get back in.
Suppose I could retroactively master a single foreign language, what language should it be? At times I would have liked to know Dutch, German, Hebrew, Japanese and the occasional French, Spanish, Danish, Italian and Portuguese. In the future I suspect I would visit countries speaking Hungarian, Russian, Chinese, Swedish and many others. I've gotten very good at navigating in countries where I don't know the language. In most European countries I can pass as a local as long as I keep my mouth shut.
The University of Chicago has a rather strict TOEFL requirement that would likely have caused a problem for me had I grown up in say Germany. Our department also has a small foreign language requirement for the Ph.D. Foreign language requirements made sense in a different era when papers were written in many languages. I remember a scene in graduate school where my advisor Mike Sipser and some Russian speaking students poured over the latest paper by Razborov translating from the Russian and hoping to understand Razborov's next great result. But now with nearly all papers written in English the requirement seems like a relic from a bygone time. Perhaps we should require every student to take the test in French, for France still has a few researchers stubborn enough to keep writing in their native tongue.
Posted by Lance to Computational Complexity at 6/1/2005 08:37:00 AM