French Fri, 16 Feb 2024 Academia ================ I studied French for five years (give or take one) in ``middle school'' (Dutch middle school combines the American middle school and high school into one, that is to say, I had French classes from age 11 till 16). I was allowed to drop French in the fifth year due to my dyslexia and would have otherwise had six years in total. I -- like most people how had any foreign language in school -- remember very little of my French classes. Most of them focussed on grammar and literature, which are two aspects I know nothing about even in my native languages. The only knowledge I retained of French were the useful phrases picked up in the first year (my name is, I don't speak French very well, can you slow down, where is the supermarket/hospital/pharmacy, etcetera). Just in case anyone was wondering, this is the same for people who dropped French after two years in favour of German, Latin, or Greek, and for people who pursued French for the full six year program. After graduating, i spent some time in France, and most of my actually useful French knowledge came from that brief period. Needless to say, when I encountered a potential source of my masters dissertation written in French, I was not exactly ecstatic about translating it. The paper seemed at first glance to be very useful however, so I made a cup of tea, grabbed my notebooks and tablet, and set off towards the study room in my village. ----------------------------------------------------------------- The paper I was translating is titled ``Valeur critique de la mystique plotienne'' by Jean Troulliard. This is a highly specialized work containing lots of Neoplatonic jargon and a fair amount of ancient Greek and Latin. Also, did I mention that `translation' was a course reserved for 6th year Frech? The year I happened to drop? Luckily, there are many tools these days to facilitate translation. Large language models are an order of magnitude more adept at translation compared to traditional digital methods such as google translate, and the availability of digital online dictionaries further facilitates the process. The first problem I had to solve however was getting the document in a machine readable format. I possessed only a scan of the document which I put through an Optical Character Recognition (OCR) program to get the paper in text from. This was actually fairly accurate. The program struggled slightly with the French quotation marks << and >>, and it completely botched the Greek, footnote markers, and occasionally interpreted a speck on the lens as a period or comma. So I went through the entire document first in French, doing an instal pass on the spelling, punctuation, and transcribing all the Greek parts by hand. I then fed this data in its entirety to Chat GPT (it is 2024 after all) and put that output into a latex document. I went thought his document as I normally do, reading and annotating important parts whist making notes in my physical notebook for the sake of redundancy. Having already made a pass on the original French, I felt as though this went a lot faster than normal. While reading, I also improved the automatic translation here and there, mostly breaking up sentence structures which appeared rather obtuse in English, and fixing some mistakes the LLM made with philosophical jargon. For this step I was assisted by someone more versed in the romance languages and French in particular, which proved not entirely necessary, but certainly improved the translation at the end. Whist reading, I noticed I was being a lot more critical. Usually when reading paper like these, I just take what the author says for granted while reading (unless it is particularly outrageous) and reflect critically at a later date. Now however, I found myself scrutinizing every line -- primarily for translation errors, but therefore also picking out lines which did not seem in-line with what I knew of the Neoplatonic tradition. At the end of the day, i spent nearly six hours reading 10 pages of text, only ~12 times slower than my usual speed. Great success. What stood out to me most about this exercise is how enjoyable and educative it was. Wanting to actually read the paper did wonders for my motivation, whereas reading French ``high literature'' in middle school was always a drag. Furthermore, I learned more about French sentence structure in these six hours than in five years of classes, and I doubt the sixth year would have helped much either. Having to transcribe the Greek by hand did wonders for my comprehension of the alphabet (which used to be shoddy at best, but is now shoddy at worst), and the whole project made me much more excited about reading French texts in the Future. ----------------------------------------------------------------- All of that is to say: (middle) school language classes (and classes in many other areas for that matter) have failed me and almost everyone I know. They both fail to actually teach, and suck so much joy out of learning that people develop permanent aversions towards certain topics. I was convinced by my teachers that I was not good at languages. Now I am fluent in two, capable in two others, and a beginner in yet another pair, whist knowing phrases and basics for a handful of languages in various families. I sincerely hope that the educational system has improved since I left, but given that its main focus currently seems to be `modernizing' -- id est: giving 4 year olds personal laptops and/or tablets, so they can stay glued to a screen even at school -- I am not expecting much.