Spell Checker & Language Translator

Started by Cogliostro, Mon 18/03/2013 17:54:56

Previous topic - Next topic

Cogliostro

I came up with a clever idea, and after a quick search, discovered it had first been posted nine years ago by Pumaman himself -- how to easily do a spell check in AGS.

Short explanation for SPELL CHECKING
1) Create a translation file.
2) Export it into MS Word, or whatever spell checker you want to use.
3) Find each typo in MS Word, switch back to AGS, do a "FIND" in "CURRENT PROJECT" for that typo and fix it.
4) Repeat step 3 until all of the typos are gone.

On the same note...

Short explanation for LANGUAGE TRANSLATION
1) Create a translation file.
2) Open it up and paste the results into Column A in EXCEL
3) Copy those lines into the Google Translator
4) Paste those results from Google into Excel in Column B, but shift down one cell.
5) In column C copy the results from the first two columns
    i.e.   [Cell C1]=A1, [Cell C2]=B2, et cetera.
6) Copy the results from Column C back into the Translation file.
7) Save.

Thought I'd bring it to the attention of those who hadn't thought of it yet.

- Cogliostro

EDIT: I just posted this to the wrong forum, could one of the moderators move it?
"First things first, but not necessarily in that order." - Dr. Who

Cerno

That's a neat workaround for getting your spelling checked.

I wondered whether spellchecking would be a nice feature to be included in AGS.
There are open source dictionaries which are basically a long list of all valid words in a given language, including all the conjugations, declensions and other stuff.
Since AGS's translation function already extracts all the text from the game scripts and definitions, that information could be fed into the dictionary and checked for correctness. Then the user would be pointed to exactly the script position where the spelling error occurred. A more sophisticated solution would include a local dictionary that contains all words that you want to ignore (game vocabulary, names, etc). The function could work similar to a word processor spell check: You have the choice to ignore, correct or add to the local dictionary.

Of course, grammatical errors would be another issue altogether. Microsoft has spent quite some money to get decent grammar checkers for different languages, which would probably difficult to re-implement.

A note about translations (although I guess it's pretty obvious): While auto-translations have gone a long way and are generally quite good, as a non-native English speaker I would have to say that playing an automatically translated game would be pretty painful and (especially in adventure games) potentially misleading since often the finer nuances are lost in the translation. Still, using the auto-translate as a basis for a proper translation could possibly save a lot of time.

Overall, nice ideas. I'll definitely use the first one for my proofreading (without the Word part) further along the road.
123  Currently working on: Sibun - Shadow of the Septemplicon

VVK

#2
Quote from: Cerno on Mon 18/03/2013 20:16:04While auto-translations have gone a long way and are generally quite good, as a non-native English speaker I would have to say that playing an automatically translated game would be pretty painful and (especially in adventure games) potentially misleading since often the finer nuances are lost in the translation. Still, using the auto-translate as a basis for a proper translation could possibly save a lot of time.
There are good machine translations now? Which ones? Google Translate certainly is still producing text that is rubbish by the standards of anything else (than machine translation) and often doesn't even preserve all of the straightforward semantic meaning. I agree that it would be painful to play a game translated with it.

I've never heard of competent translators using machine translations as a basis, and, as someone who does translation as sort of a hobby (actually I'm expecting to get a minor paid translation job in the near future too), I think I can guess why. A garbled version in your own language isn't easier to work with than the original in a language you understand. If the computer can get it right, it's probably trivial, and otherwise it's just better not to look at too literal translations lest they confuse you. I guess the thing is... real translation is anything but mechanic, you always have to think about it.

Machine translations, that I know of, are sort of passable for understanding a piece of text, but not fit to create a finished text of any sort. The method outlined above using Google Translate (well, Translator... presumably the same thing?) is technically clever, but I wouldn't use it in practice.

Snarky

Yeah, I would only use Google Translate if I didn't know the source language (like, if I had to translate a game from Greek to English; though given the lack of Unicode support in AGS, that's not the best example). But then I probably wouldn't be the right person to do the translation anyway.

At work my colleagues sometimes have to translate interview transcripts from German to English, and they tried to use Google Translate as a first pass. It did not speed things up at all; it took more time to edit the results into proper English than it would have taken them to just write the translations themselves in the first place. And German-English is one of the best-supported language pairs. (Admittedly, machine translators have particular difficulty with transcripts from informal speaking situations, where you get a lot of sentence fragments and incomplete thoughts, self-corrections, run-ons and waffling, slang, umms, aahs and grunts, omissions and non-spoken references. But still.)

I'm not at all anti computer translation, but there are some tasks it's right for and some tasks it is not.

Cerno

#4
Quote from: VVK on Mon 18/03/2013 21:36:59
I've never heard of competent translators using machine translations as a basis, and, as someone who does translation as sort of a hobby (actually I'm expecting to get a minor paid translation job in the near future too), I think I can guess why. A garbled version in your own language isn't easier to work with than the original in a language you understand. If the computer can get it right, it's probably trivial, and otherwise it's just better not to look at too literal translations lest they confuse you. I guess the thing is... real translation is anything but mechanic, you always have to think about it.

I've done some translation work myself (writing German subs for a movie as a hobby project), and I would agree that for the skilled translator, having a pre-translation would probably not help much.

However, as Snarky commented, it might help if you don't know much about the source language. Imagine for instance you're writing a game with limited means and are happy to have found a native speaker of the target language who is willing to help but is not very fluent in the source language. I guess in that use case it might be a little help.

On the other hand: I just entered a random Wikipedia article into Google Translate and you are right: The results are pretty bad, to the point where the translation is barely understandable. I guess I had some positive recollection of its capabilities because it's somewhat helpful when you absolutely have no idea about the source language.

Edit: By the way, I found this quite interesting (sorry for the off-topic):
http://www.t3.com/news/microsoft-demos-live-english-to-chinese-translation
Although I have no clue about Chinese so I can't assess the quality of the translation. The crowd seemed pleased though.
123  Currently working on: Sibun - Shadow of the Septemplicon

Cogliostro


The main idea was the spell check.

Google Translator is nothing great, however, in a game with 3000 lines most of them are along the lines of "The pencil is yellow" and Google can translate that just fine.  The villain's monologue, the exposition dump, the techno-babble, et cetera, contain considerably more complexity and having a translator is preferable.

- Cogliostro
"First things first, but not necessarily in that order." - Dr. Who

SMF spam blocked by CleanTalk