Last summer, the New York Times published a beautiful article by Sophie Hughes about translation. Hughes takes the reader through her translation process, showing how she considers nuances in tone and character perspective to iteratively translate Spanish into idiomatic English. She calls translation “a playful pursuit of equilibrium”. It’s a pleasure to follow her as she tries out and discards different translation options. If you haven’t read the piece yet, I highly recommend you do. I’ll wait.
Don't throw shade on poor AI just because it's down with the kids!
It's funny, I was just watching American Fiction, which takes on this same trope: words like "contemporary" and "real" these days are widely understood to mean something like 'about poverty and crime in a racialised way'... Perhaps the real challenge is finding what prompt you might use to get ChatGPT to talk like a normal 21st century human being.
I don't have access to the NYT, so I'm not sure what process Hughes recommends, but when I translate, I like to let myself diverge from the text for a few rounds, to free associate and brainstorm and drag in all the ideas I can, then converge back to the source text for a few rounds of editing. First you've got to let the text build its connections with everything else that you know in the target language within your rough drafts and your own mind; then you cut that all back, interrogate each word to see if you can justify it from the source text. You have to just trust that the things you found in the text will continue to be there for alert readers.
I guess what I'm really saying is that you've gotta let those texts vibe.
Fascinating, funny and a bit frightening...but AI was already frightening in all aspects. I fear we are in the process of seceding our role as reasoning and problem solving beings! Before very long in the scheme of time, machines will be the arbiters of all questions and the final authority for all answers. Yes chicken little, the sky is falling but don't ask ChatGBT...
That last version! Maybe there’s a market for comedy based on ChatGPT being bad at translating? I’d watch an AI performance of a passage or two 😆
Wait so… what was your ex’s translation?
Don't throw shade on poor AI just because it's down with the kids!
It's funny, I was just watching American Fiction, which takes on this same trope: words like "contemporary" and "real" these days are widely understood to mean something like 'about poverty and crime in a racialised way'... Perhaps the real challenge is finding what prompt you might use to get ChatGPT to talk like a normal 21st century human being.
I don't have access to the NYT, so I'm not sure what process Hughes recommends, but when I translate, I like to let myself diverge from the text for a few rounds, to free associate and brainstorm and drag in all the ideas I can, then converge back to the source text for a few rounds of editing. First you've got to let the text build its connections with everything else that you know in the target language within your rough drafts and your own mind; then you cut that all back, interrogate each word to see if you can justify it from the source text. You have to just trust that the things you found in the text will continue to be there for alert readers.
I guess what I'm really saying is that you've gotta let those texts vibe.
Fascinating, funny and a bit frightening...but AI was already frightening in all aspects. I fear we are in the process of seceding our role as reasoning and problem solving beings! Before very long in the scheme of time, machines will be the arbiters of all questions and the final authority for all answers. Yes chicken little, the sky is falling but don't ask ChatGBT...