Translation workflow

Moving this discussion to the main developer list as well as the
documentation list to ensure input from all concerned.

Thanks to the team currently contributing to a French translation of
the DHIS manual and inline help. Jason and I have been working on find
the best workflow for this work, but it is not obvious that there is
really a good one, because there are a number of issues that may be
conflicting.

Ideally, we would like to:

1) keep the original English documentation in DocBook XML format,
which allows us to generate many formats, such as PDF and HTML, as
well as incorporate the text into the online help inside DHIS 2
itself.

2) keep both the DocBook structure and translated text (which can be
in PO files) under some sort of version control, so we can track
changes, and readily generate artefacts such as PDFs

3) make it easy for translators, so they can concentrate on the
language rather than tools like Bazaar, Maven, and DocBook/XML

3b) I thought it would be easier to start from a machine translation
and do corrections, rather than start with just the English and write
everything. I would very much like the opinion of people who have some
actual experience correcting Google translations to French (the
quality is likely to differ for each language). A third possibility
would be that you just used Google translate or something similar
yourself by copy/paste, I guess.

4) easily share translations, and especially have a common and
consistently used glossary for DHIS terms such as organisation unit
and data element.

5) have an easy way of identifying which parts have changed in the
original, so the translations can be updated.

A translator could work directly with DocBook, using and XML editor
like Serna, but then machine translation becomes more tricky. For PO
files, you can use a PO editor. They can be generated and re-inserted
into DocBook (though there are some minor issues). Launchpad also
supports the PO format. For raw text, you can use Google Docs or any
editor but the issue is then how to get the translated strings back
into the DocBook structure.

And throughout, the problem of changes in the original presents
itself. To my mind, most parts of the manual will quickly
stabilize.The alternative would be to almost completely decouple the
different languages, and just keep the chapter or section headings,
sort of like Wikipedia does for different languages.

Appreciate your input - for example on which of the considerations
above you find most important.

Knut

Hi Knut,
Thanks for raising this topic. I have tried to get the french translation
done with the help of Lars using the PO Editor, but I gave up because of
problems with commit process. But this is a technical problem that can be
solved with little work. I found the PO Editor ok and quite easy to use.
And maybe it is the best way to go.
The problem here for me is organizational. Because for instance I know
that Edem also is working on French translation too, so we need to
synchronize our work to avoid duplication of efforts. I think we first need
to create translation teams by languages and select a head for each team.
Then we must provide a timeline to get the job done. Each team will
organise itself to meet the goal in the requested time. The mean they will
use (machine or human or both) is not in that case important if the team
can ensure that they are ok with the translation provided at the end of the
work.

Hope this contribution is helpful...

Regards,

Romain

···

On Wed, 15 Sep 2010 09:34:44 +0200, Knut Staring <knutst@gmail.com> wrote:

Moving this discussion to the main developer list as well as the
documentation list to ensure input from all concerned.

Thanks to the team currently contributing to a French translation of
the DHIS manual and inline help. Jason and I have been working on find
the best workflow for this work, but it is not obvious that there is
really a good one, because there are a number of issues that may be
conflicting.

Ideally, we would like to:

1) keep the original English documentation in DocBook XML format,
which allows us to generate many formats, such as PDF and HTML, as
well as incorporate the text into the online help inside DHIS 2
itself.

2) keep both the DocBook structure and translated text (which can be
in PO files) under some sort of version control, so we can track
changes, and readily generate artefacts such as PDFs

3) make it easy for translators, so they can concentrate on the
language rather than tools like Bazaar, Maven, and DocBook/XML

3b) I thought it would be easier to start from a machine translation
and do corrections, rather than start with just the English and write
everything. I would very much like the opinion of people who have some
actual experience correcting Google translations to French (the
quality is likely to differ for each language). A third possibility
would be that you just used Google translate or something similar
yourself by copy/paste, I guess.

4) easily share translations, and especially have a common and
consistently used glossary for DHIS terms such as organisation unit
and data element.

5) have an easy way of identifying which parts have changed in the
original, so the translations can be updated.

A translator could work directly with DocBook, using and XML editor
like Serna, but then machine translation becomes more tricky. For PO
files, you can use a PO editor. They can be generated and re-inserted
into DocBook (though there are some minor issues). Launchpad also
supports the PO format. For raw text, you can use Google Docs or any
editor but the issue is then how to get the translated strings back
into the DocBook structure.

And throughout, the problem of changes in the original presents
itself. To my mind, most parts of the manual will quickly
stabilize.The alternative would be to almost completely decouple the
different languages, and just keep the chapter or section headings,
sort of like Wikipedia does for different languages.

Appreciate your input - for example on which of the considerations
above you find most important.

Knut

_______________________________________________
Mailing list: https://launchpad.net/~dhis2-devs
Post to : dhis2-devs@lists.launchpad.net
Unsubscribe : https://launchpad.net/~dhis2-devs
More help : https://help.launchpad.net/ListHelp

Hi Romain

Thanks for this debate

May I know how many languages do we need to translate this manuals

Am in Tanzania is Swahili needed

If so am ready to work on it

Hope to hear from you

Regards

Japhet Kamala,

University Computing Centre Ltd

Tel: +255 22 2410645

Fax: +255 22 2410690

Cell +255 717 001807

   +255 787 123115

E-mail japhetkamala@uccmail.co.tz

       japhetkamala@yahoo.co.uk
···

From: "romain@tohouri.com" romain@tohouri.com
To: Knut Staring knutst@gmail.com
Cc: philippe philippe@veltsos.com; DHIS 2 developers dhis2-devs@lists.launchpad.net; Johan Lemarchand johan.lemarchand@gmail.com; narodar narodar@gmail.com; dhis2-documenters dhis2-documenters@lists.launchpad.net
Sent: Wed, 15 September, 2010 14:05:35
Subject: Re: [Dhis2-documenters] [Dhis2-devs] Translation workflow

Hi Knut,
Thanks for raising this topic. I have tried to get the french translation
done with the help of Lars using the PO Editor, but I gave up because of
problems with commit process. But this is a technical problem that can be
solved with little work. I found the PO Editor ok and quite easy to use.
And maybe it is the best way to go.
The problem here for me is organizational. Because for instance I know
that Edem also is working on French translation too, so we need to
synchronize our work to avoid duplication of efforts. I think we first need
to create translation teams by languages and select a head for each team.
Then we must provide a timeline to get the job done. Each team will
organise itself to meet the goal in the requested time. The mean they will
use (machine or human or both) is not in that case important if the team
can ensure that they are ok with the translation provided at the end of the
work.

Hope this contribution is helpful…

Regards,

Romain

On Wed, 15 Sep 2010 09:34:44 +0200, Knut Staring knutst@gmail.com wrote:

Moving this discussion to the main developer list as well as the
documentation list to ensure input from all concerned.

Thanks to the team currently contributing to a French translation of
the DHIS manual and inline help. Jason and I have been working on find
the best workflow for this work, but it is not obvious that there is
really a good one, because there are a number of issues that may be
conflicting.

Ideally, we would like to:

  1. keep the original English documentation in DocBook XML format,
    which allows us to generate many formats, such as PDF and HTML, as
    well as incorporate the text into the online help inside DHIS 2
    itself.

  2. keep both the DocBook structure and translated text (which can be
    in PO files) under some sort of version control, so we can track
    changes, and readily generate artefacts such as PDFs

  3. make it easy for translators, so they can concentrate on the
    language rather than tools like Bazaar, Maven, and DocBook/XML

3b) I thought it would be easier to start from a machine translation
and do corrections, rather than start with just the English and write
everything. I would very much like the opinion of people who have some
actual experience correcting Google translations to French (the
quality is likely to differ for each language). A third possibility
would be that you just used Google translate or something similar
yourself by copy/paste, I guess.

  1. easily share translations, and especially have a common and
    consistently used glossary for DHIS terms such as organisation unit
    and data element.

  2. have an easy way of identifying which parts have changed in the
    original, so the translations can be updated.

A translator could work directly with DocBook, using and XML editor
like Serna, but then machine translation becomes more tricky. For PO
files, you can use a PO editor. They can be generated and re-inserted
into DocBook (though there are some minor issues). Launchpad also
supports the PO format. For raw text, you can use Google Docs or any
editor but the issue is then how to get the translated strings back
into the DocBook structure.

And throughout, the problem of changes in the original presents
itself. To my mind, most parts of the manual will quickly
stabilize.The alternative would be to almost completely decouple the
different languages, and just keep the chapter or section headings,
sort of like Wikipedia does for different languages.

Appreciate your input - for example on which of the considerations
above you find most important.

Knut


Mailing list: https://launchpad.net/~dhis2-devs
Post to : dhis2-devs@lists.launchpad.net
Unsubscribe : https://launchpad.net/~dhis2-devs
More help : https://help.launchpad.net/ListHelp


Mailing list: https://launchpad.net/~dhis2-documenters
Post to : dhis2-documenters@lists.launchpad.net
Unsubscribe : https://launchpad.net/~dhis2-documenters
More help : https://help.launchpad.net/ListHelp

Hi Romain,

This is part of the reason that we created the workflow using Serna,
brz and Launchpad. We have a set of distributed individuals that are
making changes, potentially at the same time, to the same document.
Version control systems are very good at handling and merging these
types of changes. However, it was felt that in order to extend the
translation of the documents to persons who were not familiar with
these tools, that PO Editor was a better choice. It is a great choice
for translators, as they only need to worry about the translation of
strings, and not anything else in the document itself. So, no need to
learn DocBook.

The use of bzr (a version control system) is still possible with PO
files themselves, but requires some familiarity with the use of the
tool and the use of Launchpad. We can consider putting these PO files
under version control, which will allow a translation team to work in
a distributed way.

The other possibility is the use of the Google Spreadsheet which has
been setup. When this is used, changes to the document is recorded
automatically, even when different users are editing the same
document.

Another possibility is the use of Launchpads own translation system,
which we need to investigate more.

As I have sort of alluded to, I still believe that the use of a
version control system is the only feasible technical way to solve
these problems. However, it does require a certain amount of learning
to get used to using the tools.

Johan, Phillipe, myself and hopefully Knut will meet week in Geneva to
discuss some of these issues. You feedback is very valuable, and it
would be good to know what you would be comfortable using a version
control system like bzr. You can consult the DHIS2 Documentation Guide
for more information about how it works.

Best regards,
Jason

···

On Wed, Sep 15, 2010 at 1:05 PM, <romain@tohouri.com> wrote:

Hi Knut,
Thanks for raising this topic. I have tried to get the french translation
done with the help of Lars using the PO Editor, but I gave up because of
problems with commit process. But this is a technical problem that can be
solved with little work. I found the PO Editor ok and quite easy to use.
And maybe it is the best way to go.
The problem here for me is organizational. Because for instance I know
that Edem also is working on French translation too, so we need to
synchronize our work to avoid duplication of efforts. I think we first need
to create translation teams by languages and select a head for each team.
Then we must provide a timeline to get the job done. Each team will
organise itself to meet the goal in the requested time. The mean they will
use (machine or human or both) is not in that case important if the team
can ensure that they are ok with the translation provided at the end of the
work.

Hope this contribution is helpful...

Regards,

Romain

On Wed, 15 Sep 2010 09:34:44 +0200, Knut Staring <knutst@gmail.com> wrote:

Moving this discussion to the main developer list as well as the
documentation list to ensure input from all concerned.

Thanks to the team currently contributing to a French translation of
the DHIS manual and inline help. Jason and I have been working on find
the best workflow for this work, but it is not obvious that there is
really a good one, because there are a number of issues that may be
conflicting.

Ideally, we would like to:

1) keep the original English documentation in DocBook XML format,
which allows us to generate many formats, such as PDF and HTML, as
well as incorporate the text into the online help inside DHIS 2
itself.

2) keep both the DocBook structure and translated text (which can be
in PO files) under some sort of version control, so we can track
changes, and readily generate artefacts such as PDFs

3) make it easy for translators, so they can concentrate on the
language rather than tools like Bazaar, Maven, and DocBook/XML

3b) I thought it would be easier to start from a machine translation
and do corrections, rather than start with just the English and write
everything. I would very much like the opinion of people who have some
actual experience correcting Google translations to French (the
quality is likely to differ for each language). A third possibility
would be that you just used Google translate or something similar
yourself by copy/paste, I guess.

4) easily share translations, and especially have a common and
consistently used glossary for DHIS terms such as organisation unit
and data element.

5) have an easy way of identifying which parts have changed in the
original, so the translations can be updated.

A translator could work directly with DocBook, using and XML editor
like Serna, but then machine translation becomes more tricky. For PO
files, you can use a PO editor. They can be generated and re-inserted
into DocBook (though there are some minor issues). Launchpad also
supports the PO format. For raw text, you can use Google Docs or any
editor but the issue is then how to get the translated strings back
into the DocBook structure.

And throughout, the problem of changes in the original presents
itself. To my mind, most parts of the manual will quickly
stabilize.The alternative would be to almost completely decouple the
different languages, and just keep the chapter or section headings,
sort of like Wikipedia does for different languages.

Appreciate your input - for example on which of the considerations
above you find most important.

Knut

_______________________________________________
Mailing list: DHIS 2 developers in Launchpad
Post to : dhis2-devs@lists.launchpad.net
Unsubscribe : DHIS 2 developers in Launchpad
More help : ListHelp - Launchpad Help

_______________________________________________
Mailing list: DHIS 2 developers in Launchpad
Post to : dhis2-devs@lists.launchpad.net
Unsubscribe : DHIS 2 developers in Launchpad
More help : ListHelp - Launchpad Help

--
Jason P. Pickering
email: jason.p.pickering@gmail.com
tel:+17069260025
sip:jason.p.pickering@ekiga.net