Re: Emacs as a translator's tool

[Eric Abrahamsen]

Marcin Borkowski <mbork@mbork.pl> writes:

> On 2021-12-21, at 18:41, Eric Abrahamsen <eric@ericabrahamsen.net> wrote:
>
>> Kevin Vigouroux via Users list for the GNU Emacs text editor
>> <help-gnu-emacs@gnu.org> writes:
>>
>>> Hello,
>>>
>>> At first, personally, I would have started from a large number of
>>> existing features that I could have reused.
>>>
>>> For the design, I would rather point towards XLIFF 2 [2], precisely
>>> because fundamental ideas have already been formulated and implemented.
>>
>> Part of the problem is that "translation" ends up meaning very different
>> things in different contexts. I mostly translate fiction or other long
>> chunks of prose, but occasionally film scripts and subtitles. The
>
> Out of curiosity: what tools do you use?  (I translate subtitles a lot,
> and I used subed-mode and mpv.)

When I do subtitles it's pretty much always for a film production
company, which you'd think would mean latest technologies and industry
standards, exciting stuff, but in fact means they email me a Word
document, and expect me to email them a Word document back :(

If I'm lucky, I then get a round of editing with them, but that means
going into the studio and sitting behind the editor and saying "that one
goes by too fast". :( :(

Translating fiction for publishing houses is the same: Word documents,
track changes. So basically I'm pasting the text into a text-mode or
org-mode buffer, doing my translation, and then exporting to odt/docx.
It's very depressing.

>> requirements there are very different (and much more fluid) than
>> something like XLIFF, which seems firmly aimed at localization. I can't
>> imagine a major-mode that would be practically useful for both
>> localization and fiction translation.
>
> Agreed.  Actually, a global minor mode might be a better fit.  It could
> support e.g. linking between two buffers ("source" and "translation")
> and glossary support.

Right. Jean-Christophe mentioned TMX, which makes a lot of sense, as its
an external storage format, and usable no matter what format your actual
translation is in. And it's good for collaborative translations, though
I don't do that. A minor mode for that would be perfect, and that's
probably what I should have written.

Then another minor mode for linking and following source and
translation, as you say, but not tied to Org. org-translate dumps
segmentation control characters into the files, an approach which I
guess is still usable for any plain-text file format. Maybe there are
better ways, though. I don't know how you would manage correspondences
between a PDF and text file, for instance.

Lastly, if I ever magically become a better programmer than I am, I
would like a minor mode that keeps the translation in a VC repository,
and lets me do multiple drafts of various sections, keeping those drafts
in version control branches. Then there's a visual indicator in the
buffer that this paragraph has multiple versions, a command to display
all those versions side-by-side, and something to choose one of them
eventually (merging the draft branch to master under the hood).

Some day!

Eric