ICanLocalize translation issues - what is non ICL work flow?

Please post all questions and comments regarding Help & Manual 7 here.

Moderators: Alexander Halser, Tim Green

Post Reply
Dave Gehman
Posts: 575
Joined: Sat Sep 23, 2017 9:05 pm

ICanLocalize translation issues - what is non ICL work flow?

Unread post by Dave Gehman »

We had thought we could use I Can Localize to translate our H+M file into Japanese. There have been technical problems and this approach may not be possible.

Does any translation service offer translation integrated with H+M files? We have found none.

Alternatively, what is the right work flow for translation outside of an integrated approach?

Send only the Topics XML files?

Once they are translated and, I assume, returned as XML files in Japanese:
Create a Japanese help project and manually cut-and-paste each topic's translation into the new project?

Would hyperlinks and referenced snippets survive this process?
User avatar
Tim Green
Site Admin
Posts: 23156
Joined: Mon Jun 24, 2002 9:11 am
Location: Bruehl, Germany
Contact:

Re: ICanLocalize translation issues - what is non ICL work f

Unread post by Tim Green »

ICL supports Help+Manual projects directly, provided you use the uncompressed HMXP format. Contact ICL support for more details.

HMXP is really the format you should be using for all ongoing work. It supports granular backups, multi-user editing, version control, translation with external, XML-based tools and unlike the single compressed binary HMXZ files, it is not possible for a single catastrophic failure in Windows or your network to trash your entire project. This last issue has been mitigated by the new automatic rescue copy system that was implemented recently, but it is still something you want to avoid.
Regards,
Tim (EC Software Documentation & User Support)

Private support:
Please do not email or PM me with private support requests -- post to the forum directly.
Andreas Boettcher
Posts: 21
Joined: Fri Jun 07, 2013 7:52 am

Re: ICanLocalize translation issues - what is non ICL work f

Unread post by Andreas Boettcher »

Hi Dave,

translating H+M projects can be quite tricky.
Just translating the topic files won't do. The hmxp itself contains translatables such as the project title and variables. The TOC xml files in the maps folder contain the toc item names. The topic files themselves can contain non-translatables such as source code, comments, conditions etc.

Every professional translation provider will process all those files with a translation software called translation memory system (TMS). That system needs to be set up for each type of file you need translated, i.e. for the hmxp, the toc, the topic files. The system extracts every piece of text that needs to be translated from a document. For this to work, it relies on filters, which tell it which text segments are relevant. The more custom your approach to H+M is, the more custom those filters need to be. Let's assume you use a style "code" and you want all text formatted with that style not to be translated. The translation company would have to adjust it's filters to exclude those segments. Another example would be to exclude all topic files which have a certain status. Or topics that have a certain keyword. Or topics that contain a trigger word in their description element. You get the idea.

After translation, the TMS exports all files. During that process, it creates a copy of the source file but inserts the translated text where the original text was. The TMS can be configured to recreate the original folder structure. Note that the translation sibling feature of H&M is completely irrelevant here. The translator will not work with H+M.

Basically, what the system does is:
- Get files from folder "source"
- Weed out the code and non-translatable bits
- Have a translator translate the relevant bits
- Create the translated documents
- Store the translated documents in folder "target"

Note that the source directory structure isn't really all that important, as long as you have all the stuff of your source language in one directory. A useful setup could look like this:

.../product/documentation/manual/en
.../product/documentation/manual/de
.../product/documentation/manual/jap
.../product/documentation/manual/fr

That way, you just paste the translated project to the target language folder. All references will still work. Depending on how thorough the TMS was configured, that's it. The target project will be a 1:1 copy of your source project, except for the bits and pieces you have told the TMS to translate.
Last edited by Andreas Boettcher on Fri Feb 23, 2018 11:03 am, edited 1 time in total.
Dave Gehman
Posts: 575
Joined: Sat Sep 23, 2017 9:05 pm

Re: ICanLocalize translation issues - what is non ICL work f

Unread post by Dave Gehman »

Tim Green wrote:ICL supports Help+Manual projects directly, provided you use the uncompressed HMXP format. Contact ICL support for more details.
Yes, that file format (HMXP) is the type we are using. There is a serious technical problem between us and ICL (and of course we are working the ICL support).

It looks as though we are going to have to proceed outside of H+M and/or ICL's direct support of H+M
Dave Gehman
Posts: 575
Joined: Sat Sep 23, 2017 9:05 pm

Re: ICanLocalize translation issues - what is non ICL work f

Unread post by Dave Gehman »

Many thanks for your thoughtful answer, Andreas
Andreas Boettcher wrote:Every professional translation provider will process all those files with a translation software called translation memory system (TMS). That system needs to be set up for each type of file you need translated, i.e. for the hmxp, the toc, the topic files. The system extracts every piece of text that needs to be translated from a document. For this to work, it relies on filters, which tell it which text segments are relevant.
How is this done? For example, one of the translators whose credentials fit the content of our project uses Trados Studio. How can she set up Trados Studio so that Studio can work most efficiently? (We aren't looking for step-by-step instructions here -- just a very high level overview of the work flow).
The more custom your approach to H+M is, the more custom those filters need to be.... After translation, the TMS exports all files. During that process, it creates a copy of the source file but inserts the translated text where the original text was. The TMS can be configured to recreate the original file structure. Note that the translation sibling feature of H&M is completely irrelevant here. The translator will not work with H+M.

Basically, what the system does is:
- Get files from folder "source"
- Weed out the code and non-translatable bits
- Have a translator translate the relevant bits
- Create the translated documents
- Store the translated documents in folder "target"

....{Further useful info around directories}....

you just paste the translated project to the target language folder. All references will still work. Depending on how thorough the TMS was configured, that's it. The target project will be a 1:1 copy of your source project, except for the bits and pieces you have told the TMS to translate.
I'm not grasping how the TMS works the magic here -- the magic being the ability simply to drop returned (translated) files into the H+M project, so that we can enjoy the automated output type (in our case, WebHelp).

Sorry to be dense, but this is our first foray into translating a complete help project in such a way that we can fully enjoy all the benefits of H+M.

H+M is proving to be a superb authoring environment... we're just puzzled about the work flow around translation.
User avatar
Tim Green
Site Admin
Posts: 23156
Joined: Mon Jun 24, 2002 9:11 am
Location: Bruehl, Germany
Contact:

Re: ICanLocalize translation issues - what is non ICL work f

Unread post by Tim Green »

Hi Dave,
I'm not grasping how the TMS works the magic here -- the magic being the ability simply to drop returned (translated) files into the H+M project, so that we can enjoy the automated output type (in our case, WebHelp).
What you get back from a translator working with translation memory is a completely new copy of the project in the target language. As with all translation in Help+Manual, each language version is in a separate project, that is and must be separate from your original project.

The advantage of translation memory is that it maintains a database of everything that has already been translated with a very high degree of granularity. When the time comes to update the translation you just send the translator the entire updated version of the original project (always the full project, never just parts). The translation memory tool then processes it automatically and replaces all known sentences with the translations that the human translator already made on the first version. All that is left is new and altered text, which the human translator then processes and translates manually, with support from translation memory for things like already known phrases and terminology. This is very efficient and it is also the best way to achieve maximum terminology consistency.
Regards,
Tim (EC Software Documentation & User Support)

Private support:
Please do not email or PM me with private support requests -- post to the forum directly.
Dave Gehman
Posts: 575
Joined: Sat Sep 23, 2017 9:05 pm

Re: ICanLocalize translation issues - what is non ICL work f

Unread post by Dave Gehman »

Tim Green wrote:What you get back from a translator working with translation memory is a completely new copy of the project in the target language.
So, if I'm understanding correctly, the TMS simply leaves untouched the support and baggage goodies in the directories for the H+M project except the text to be translated. Then the TMS creates a new directory structure that copies in the non-translated support files + the translated Topics files. That new directory structure mirrors the original H+M directory, and it is this new directory that becomes the H+M_Japanese (for example) project.

Correct?
User avatar
Tim Green
Site Admin
Posts: 23156
Joined: Mon Jun 24, 2002 9:11 am
Location: Bruehl, Germany
Contact:

Re: ICanLocalize translation issues - what is non ICL work f

Unread post by Tim Green »

Hi Dave,
Correct?
No, you're over-thinking this. It doesn't need to create or re-create or restructure anything. The structure is already there and doesn't get touched, or need to get touched. It works directly on the XML files. It edits your HM project directly, just as if you were working on it directly in HM. That is the beauty of XML in the uncompressed state. What you get back is the original project, just as you sent it out, just with the text translated.
Regards,
Tim (EC Software Documentation & User Support)

Private support:
Please do not email or PM me with private support requests -- post to the forum directly.
Dave Gehman
Posts: 575
Joined: Sat Sep 23, 2017 9:05 pm

Re: ICanLocalize translation issues - what is non ICL work f

Unread post by Dave Gehman »

Thanks again - very much appreciated.
Post Reply