Добавлено: Пт Окт 31, 2014 1:19
Заголовок сообщения:
I've been busy today and managed to get a lot done. I have now finished aligning all relevant 13.0421 files in the Text and var folders and managed to extract
25284 unique Translation Units into a TM. TU's contain both languages and are sometimes as small as one word (i.e. the prefixes) and sometimes contain a very long sentence (i.e. the dialogs).
I also created a so-called Termbase with over
3567 unique terms in it. This is a database of terms, rather than sentences, so mostly units of one to about three words. I loaded all of the names into that, the ones in the long list Boon947 kindly provided, and the ones from versions later than 13.0421 from Drakon's list of names. A termbase is used slightly differently from a TM. A TM checks and, when possible, automatically translates whole sentences. A termbase just flags and displays all separate terms with their translations that occur in the current sentence. This means that if a new sentence, which is not translated automatically, but needs to be translated manually or by Google, has words in it that have already been translated and occur in the termbase, I will see that at a glance and can quickly improve the Google translation that way, using proper terminology. There are even more nifty functions, like auto-suggestion while editing, which I won't go into now, but is another cool feature.
So after all of that was set up and prepared, I loaded the Russian Text- and var-files of version 14.0920 into Trados, at least the ones that have English language equivalents.
On that topic, there seem to be a number of files in the var folder that were not translated into English in version 13.0421, and still have cyrillic Russian in them. They are:
encounter, event, group_coord, guard, landmark, mapobject and yuplay
Do I indeed leave these as is, also in the English version? This seems a bit strange...
There are others which don't have any language components in them at all, just characters and numbers. They are:
camp_event, CampDiff, castle, heroPrice, mercenary, stat, stat_progress and terr_layer
I guess the same applies, just copy them over, no need to translate?
All files of 13.0421 take up 4.1 MB while 14.0920 is 4.91 MB, so quite a bit more content. This is reflected in the stats I got in the Trados analysis:
Total no. of segments:
240043
Total no. of words:
528784
75% of those words are 100% matches or repetitions, which means they'll get automatically translated!
13% are matching more than 70%, which means that the relevant TU will come up, but only 30% of the sentence or less needs to be edited.
12% are new sentences, lines or terms that still need to be translated, but a part of those segments may also just contain numbers and can just be copied over.
Only the "real" new sentences need to be properly translated, but Google will do that automatically in those cases, so with the help of the termbase and auto-suggestion dictionaries, this can go relatively fast and for the most part it should be possible to quickly improve those Google results and create translations of high enough quality. I estimate that only in very few cases will I need some input from Russian speakers. I'll leave those bits open for now and place them on a separate semi-public Google sheet. I don't wish to publish the link in a post, but if you know Russian and English and have the time and desire to support this process, please send me a PM and I'll mail it to you.
That's it for today, now the actual translation will start. I'll keep you posted...
The Guild of Craftsmen
