Tom Gally <tomg...@gmail.com>: Oct 05 05:54PM +0900
A followup to my initial post to this thread.
I just translated a short speech to be given at an international meeting.
The original Japanese was dense and formal, and I wasn’t happy with my
translation at first—it was accurate and correct, but I was afraid it might
be difficult for the speaker to read aloud and for the listeners to
understand. So I went to my AI editors again, running the original and my
translation successively through ChatGPT, Claude, and Gemini and asking for
their comments and suggestions. My prompt was similar to the one I used
before. Once again, some of their suggestions were quite good, and the
final version was better than what I had produced on my own.
As noted earlier in this thread, one advantage of AI editors is that I
don’t get annoyed by their suggestions and corrections. Another, I realized
now, is that I don’t have to worry about annoying *them* when I have
them comment on slightly modified versions of the same text again and again.
Another thing to note is that, while these AI editors can provide good
feedback on accuracy, meaning, and style, they are not the best
proofreaders—they sometimes failed to catch my typos and dropped
prepositions. I spotted those mistakes either when I pasted the translation
back into Word and it underlined the errors or when I had one of OpenAI’s
voices read aloud the translation for my final check.
Tom Gally
|