We all know the feeling, you’ve found a reference to a newspaper article from 1920s Paris, you’ve tracked it down in Gallica, and finally you’re ready to learn about the Parisian musical landscape. As the scan slowly loads in you let out a sigh of despair, “ah beans, you can’t actually read French. As preposterous as it is that old French newspapers didn’t bother translating themselves into English for our convenience, it is the unfortunate world we live in. What is one to do when faced with such a challenge? Of course asking someone to translate it for you works, however when you have many long articles to translate this isn’t feasible unless you have an extremely large number of French speaking friends. So, weary and broken, we turn to the devil of shoddy translation, Google translate.
(cue thunder and lightning)
I have seen Google translate before in language classes, often with… interesting results. It can be fun to see what happens when translating a text to another language and then back. And yet, it’s actually a really useful tool for research. The difference is that language classes focus on the details of grammar, which is useful to become fluent in a language but isn’t necessary for understanding an article. One could imagine being able to piece together meaning even if just given definitions of each word. Google translate isn’t a perfect tool, but it doesn’t have to be. And I think that its perception has been unfairly maligned by using it in the wrong way. So what might be a better use case? Let’s dig out that old French newspaper again. We’ve found what appears to be the right section, something tells me I can find music related things on the page labeled “les arts”, and I think I can also figure out what “theatres” and “music-halls” translate to. In this case the newspaper is an image file so we can’t directly copy text, and the Gallica OCR can’t show where the text came from in the document. Fortunately, Google Lens can take an image from your camera, grab the text, and then translate it. OCR certainly doesn’t make translation more accurate, but it’s good enough to find the “today” section where all the concerts are listed. Machine translation isn’t bad or good, but it’s important to understand its applications and limits. The best translation is the one you can actually get, even if it kinda sucks.