User Interface Localization / Language Codes

[Sabine Emmy Eller]

Actually these are two connected questions or better various points.

1) As to the UI localization: is the source file a language.properties file? If so: it can be directly translated using OmegaT. This would avoid the conversion. Furthermore: translators are not used to do by "themselves" that is, copy from the page etc. So a file should be ready to be downloaded and translated instead of having to create an own one by copying and pasting and then adding a new language. This is just a hint - I know it is not difficult to do a copy/paste, but I also know my colleagues.

Their workflow is:
*take source file
*load it in CAT-Software
*translate
*export to target

1.1) How are the translations and updates maintained?

2) Like mentioned in the mail: IMHO it would make sense to have "UI-Translation-days" going on around International Mother Language Day on 21 st of February. We will have a translation marathon for Dictionary entries.
http://www.voxhumanitatis.org/content/international-mother-language-day-2010
we also have a facebook group for that:
http://www.facebook.com/profile.php?id=682271047&ref=ts#!/group.php?gid=209376066111&ref=ts

3) Language codes: your page is linked to ISO 639-2 langauge codes which in the meantime anyway is not enough because it leaves out over 6000 languages existing in the world. Is there a possibility to support ISO639-3 language codes?

I think that's all for now.

Cheers, Sabine

[Gert]

Sabine,

1.

1) As to the UI localization: is the source file a language.properties file?

The file DictionaryForMIDs.languages contains the text items (labels) that are displayed on the screen, as well as the translations for each language. See http://dictionarymid.sourceforge.net/LanguageUITranslations.html for some more information.

Furthermore: translators are not used to do by "themselves" that is, copy from the page etc. So a file should be ready to be downloaded and translated instead of having to create an own one by copying and pasting and then adding a new language. This is just a hint - I know it is not difficult to do a copy/paste, but I also know my colleagues.

Thanks for giving us insight in the way of working for a translator !

Handling the file DictionaryForMIDs.languages is not comfortable, I know this. In order to add a language, a section needs to be added at the end of DictionaryForMIDs.languages and each line of that section must contain the translation for the corresponding text item. Likely a translator will use an intermediate step with an Excel file or similar; say what is CAT-Software ?

One more hint: translating only the English words to another language may not always achieve the best result, cause the context may not be obvious; the section UIDisplayTextItems provides some more information on the context.

Ah, if there is any question or problem, please do ask.

1.1) How are the translations and updates maintained?

Via updates of the file DictionaryForMIDs.languages. DictionaryForMIDs.languages is put under 'revision control' in the DfM SVN repository at Sourceforge (I do put the file in the SVN repository).

The file DictionaryForMIDs.languages is parsed by the tool LanguageUIGeneration, see http://dictionarymid.sourceforge.net/LanguageUIGeneration.html (I do run LanguageUIGeneration after people provided the updated file DictionaryForMIDs.languages to me).

2.

2) Like mentioned in the mail: IMHO it would make sense to have "UI-Translation-days" going on around International Mother Language Day on 21 st of February. We will have a translation marathon for Dictionary entries.

Oh, that would be great to get support for the UI translations for DfM !! (a) For getting additional languages translated - and (b) to get existing languages completed. Concerning (b): we have the problem that at some point of time a person provides the UI translation for a language, but then the DfM development advances and we amend text items in the file DictionaryForMIDs.languages, for which we then do not have the translations (English is then used as a fallback for the missing text items). So for some UI translations now there are a few missing translations (it is roughly 3 to 10 missing text items for some languages).

Maybe ... you simple could try to circulate the file DictionaryForMIDs.languages at some persons who like to do such a sort of translation work ? I guess your organization and the International Mother Language Day really could be a chance for DfM to get more and better UI translations !


3.

your page is linked to ISO 639-2 langauge codes

Um, what exactly do you mean there ? The reference to ISO 639 for the parameter languageXFilePostfix ?

Best regards,
Gert

[Gert]

Oh, one thing I forgot above: UI translations are currently only implemented for the Java ME version. Achim's Android version does not yet support the UI translations (unless Achim already implemented that recently).

Gert

[Sabine Emmy Eller]

-->Hi :-)

-->First of all let me see if the layout will be correct - anyway my answers will start with -->

Furthermore: translators are not used to do by "themselves" that is, copy from the page etc. So a file should be ready to be downloaded and translated instead of having to create an own one by copying and pasting and then adding a new language. This is just a hint - I know it is not difficult to do a copy/paste, but I also know my colleagues.

Thanks for giving us insight in the way of working for a translator !

Handling the file DictionaryForMIDs.languages is not comfortable, I know this. In order to add a language, a section needs to be added at the end of DictionaryForMIDs.languages and each line of that section must contain the translation for the corresponding text item. Likely a translator will use an intermediate step with an Excel file or similar; say what is CAT-Software ?

--> no, Excel files are not processed properly in most CAT-Tools. In case: a simple .txt file is fine. Normally utf-8 should be a good option since Windows and Linux do understand it well. Another good possibility is .po format where you can add information to each segment to be translated. This helps the translator to understand what he/she is actually translating.

--> CAT = Computer assisted translation - there are three tools I am mainly working with:
--> Omegat: http://omegat.org
--> Anaphraseus: http://anaphraseus.sourceforge.net/
--> Virtaal: http://virtaal.org

--> OmegaT is the one with the biggest community. Most people in their discussion group on yahoogroups deal with Open Source and could potentially help to localise the software.

--> A CAT-Tool helps you to create a TM (Translation Memory) which is a file in XML format. When you then have updates to a specific project you just load the old TM and you get all that was already translated as "100% match", eventually partial matches if there were changes in some segments or you have to translate what is completely new. This makes the localization process much faster.

One more hint: translating only the English words to another language may not always achieve the best result, cause the context may not be obvious; the section UIDisplayTextItems provides some more information on the context.

--> Like I said above: .po format should be a viable option then. Another possibility could be including the translations in Ambaradan and have people work there, but for that I have to talk to the developer first, since at the very basis Ambaradan is a dictionary application and I am actually not sure if we want to see software localization in there, even if it seems quite interesting to test it with such a small UI.

Ah, if there is any question or problem, please do ask.

--> Yes, I will :-)

1.1) How are the translations and updates maintained?

Via updates of the file DictionaryForMIDs.languages. DictionaryForMIDs.languages is put under 'revision control' in the DfM SVN repository at Sourceforge (I do put the file in the SVN repository).

The file DictionaryForMIDs.languages is parsed by the tool LanguageUIGeneration, see http://dictionarymid.sourceforge.net/LanguageUIGeneration.html (I do run LanguageUIGeneration after people provided the updated file DictionaryForMIDs.languages to me).

--> This means you anyway have to do this step manually. Ok - so let's see what could work. First I have to understand better what is already there, what is needed and how to combine things.

2.

2) Like mentioned in the mail: IMHO it would make sense to have "UI-Translation-days" going on around International Mother Language Day on 21 st of February. We will have a translation marathon for Dictionary entries.

Oh, that would be great to get support for the UI translations for DfM !! (a) For getting additional languages translated - and (b) to get existing languages completed. Concerning (b): we have the problem that at some point of time a person provides the UI translation for a language, but then the DfM development advances and we amend text items in the file DictionaryForMIDs.languages, for which we then do not have the translations (English is then used as a fallback for the missing text items). So for some UI translations now there are a few missing translations (it is roughly 3 to 10 missing text items for some languages).

Maybe ... you simple could try to circulate the file DictionaryForMIDs.languages at some persons who like to do such a sort of translation work ? I guess your organization and the International Mother Language Day really could be a chance for DfM to get more and better UI translations !

--> Simple question: do you want to create a "DfM localisation marathon" for IMLD 2010? Will get you a description by mail - it would be come too much here - only giving the final text/link should be less confusing for readers.


3.

your page is linked to ISO 639-2 langauge codes

Um, what exactly do you mean there ? The reference to ISO 639 for the parameter languageXFilePostfix ?

--> In the file to be localised there is a link to a language code list: http://ftp.ics.uci.edu/pub/ietf/http/related/iso639.txt - this is a very short list and it excludes the majority of languages. We use ISO 639-3 for all languace codes which is maintained by SIL http://sil.org/iso639-3/codes.asp

--> Sorry but this got very long, but cutting out your comments would make it more difficult for others to follow.

Cheers, Sabine

[Sabine Emmy Eller]

This is the latest UI file, right?

http://dictionarymid.cvs.sourceforge.net/*checkout*/dictionarymid/DictionaryForMIDs/src/de/kugihan/dictionaryformids/hmi_java_me/uidisplaytext/DictionaryForMIDs.languages

One thing that is not clearly visible here is what is "missing" in the translated versions.

[axin]

Achim's Android version does not yet support the UI translations (unless Achim already implemented that recently).

The Android version supports localization from its beginning, I just did not advertise this as I was expecting many ongoing changes.
It's basically stable now, check out the XML-files here:
view
download

Cheers,
Achim

UPDATE: links point to trunk version

[Gert]

@Sabine:

http://dictionarymid.cvs.sourceforge.net/*checkout*/dictionarymid/DictionaryForMIDs/src/de/kugihan/dictionaryformids/hmi_java_me/uidisplaytext/DictionaryForMIDs.languages

NO ! Good that you asked ! This link points to the old CVS repostitory, and I thought that one is deactivated. Here is the correct link to the SVN repository: http://dictionarymid.svn.sourceforge.net/viewvc/dictionarymid/trunk/JavaME/src/de/kugihan/dictionaryformids/hmi_java_me/uidisplaytext/DictionaryForMIDs.languages?view=markup (I just updated the wrong link on the web page - thank you for asking !!)
I believe the only major changes between these two files is that the Norwegian language was added.

One thing that is not clearly visible here is what is "missing" in the translated versions.

Yes, that is true ! The missing words are missing lines at the end of the section for a language. If you put line by line of the language section next to the section UIDisplayTextItems (or the English section) then you will detect the missing words. I know that this is very inconvenient.

--> CAT = Computer assisted translation - there are three tools I am mainly working with:

Thank you very much for your explanations on the CAT-SW. I did not yet make any experience with such as SW.

--> Simple question: do you want to create a "DfM localisation marathon" for IMLD 2010? Will get you a description by mail - it would be come too much here - only giving the final text/link should be less confusing for readers.

I would like get into such a "DfM localisation marathon" - also for the Android version (see the posting of Achim). Here is the problem: I am restriced from some time next week through mid of March cause of my job. I hardly will be able to provide any support during that time. Hmmm, maybe Achim could ... ?

--> In the file to be localised there is a link to a language code list: http://ftp.ics.uci.edu/pub/ietf/http/related/iso639.txt - this is a very short list and it excludes the majority of languages. We use ISO 639-3 for all languace codes which is maintained by SIL http://sil.org/iso639-3/codes.asp

I need to check that: the devices will report a 2-letter code in microedition.locale; I need to read the CLDC/MIDP spec there.


@Achim: thanks for your clarification ! I guess best we look for UI localizations for the Android versions also, right ? Would you have any time to support this ... also for the Java ME version ?


Thanks a lot to both of you !!
Gert

[Gert]

@Sabine:
From JSR-118 (MIDP-2.0)

The language codes MUST be the lower-case, two-letter codes as defined by ISO-639.

But we can use 3-letter codes for those languages where there is no 2-letter-code.

Best regards,
Gert

[Sabine Emmy Eller]

But we can use 3-letter codes for those languages where there is no 2-letter-code.

Ok, this means the headers of the columns should be adapted.

[Sabine Emmy Eller]

Achim's Android version does not yet support the UI translations (unless Achim already implemented that recently).

The Android version supports localization from its beginning, I just did not advertise this as I was expecting many ongoing changes.
It's basically stable now, check out the XML-files here:
view
download

Hmmm ... this means that actually we have "two projects" with different formats, txt and xml. One person could do all since partly translations will be the same and there are not too many strings. I have some ideas, but have to test this. I'll let you know about it.

Is there any further stuff that would be relevant in many languages? Something like this documentation: http://dictionarymid.sourceforge.net/instruction.html ? How up to date is it (well I should see it this afternoon).

Sabine

[Gert]

Hmmm ... this means that actually we have "two projects" with different formats, txt and xml. One person could do all since partly translations will be the same and there are not too many strings. I have some ideas, but have to test this. I'll let you know about it.

Wow - great !!

Is there any further stuff that would be relevant in many languages? Something like this documentation: http://dictionarymid.sourceforge.net/instruction.html ? How up to date is it (well I should see it this afternoon).

Oh - that instruction is probably 5 years old 

Best regards,
Gert