Start a new topic

Improve auto-assembling with machine translation

Starting to work with a new source language (English) I'm discovering the virtues of the feature Improve auto-assembling with machine translation:


image


It's impressive to see how the results in the three MT panes (Google Translate, Microsoft Translator and Yandex) are updated on the fly when I add new terms to the glossary:


image


MT results updated as:


image


However, since there always must be a 'however':

  • Why is this colour-coding green and not orange (for glossary entries)?
  • It would still be useful to see the original, not updated MT results, e.g. by hovering over them or by placing the original result below the updated (improved) MT result. Would such an improvement be possible?


> It would still be useful to see the original, not updated MT results, e.g. by hovering over them or by placing the original result below the updated (improved) MT result. Would such an improvement be possible?


I don't know it today. I may have eureka tomorrow. 


> Why is this colour-coding green and not orange (for glossary entries)?


One color for MT result improvements (no matter with TM fragments or glossary entries) is really enough.

>One color for MT result improvements (no matter with TM fragments or glossary entries) is really enough.


True, but consistent colour coding would suggest a new colour ;) ?

I've always been confused by the name of this feature.
Reading the above, it appears to actually be "Improve machine translation with auto-assembling", i.e. the MT results are modified by glossary entries and other AA features, and not vice versa as the name implies.
Is that correct?
If so, please rename this so that it makes sense. I'd have been using this feature for the past two years (and would have got more value out of CT) if it was properly named.

Or:


Improve machine translation via auto-assembling

I've used this improvement now for 2 days and I've found a mayor drawback for me: Engines like Google Translate can give valuable clues for words. However, sometimes these useful target terms are only shown in specific sentences, based on context.


So, in segment 23 a source word is translated by Google Translate with a specific target term, and in segment 456 a far better target term is used: one that I'd like to add to my glossary and that I'd like to insert via F/R in the TM and previously translated segments.


The thing is, that with:


image


active, my previously used, suboptimal target term will always be inserted and displayed:


image


Only when I click 'Translate selected fragment', I'll see Google Translate's new and better translation for the target term.


So here's what I'd like to suggest:


  • Turn the mechanism around: Even with the improvement active, no target terms will be inserted into the Google Translate translation.
  • Rename 'Translate selected fragment' to 'Update translation via AA' when improvement is active.
  • Only by clicking on 'Update translation via AA' the Google Translate result with the injected AA results, based on your own glossaries and memories, will be shown.
  • If you're happy with it, click 'Transfer into target segment' or use the keyboard shortcut.

Login to post a comment