MB fund translation of Musicbrainz into one of remaining untranslated top 10 global languages?

[mmirG]

There are major languages in which huge numbers of music recordings exist but Musicb&Picard are not available in those languages.

An experiment in allocating MB funds to pay for the translation of Musicbrainz and Picard into one of these languages would be illuminating.

Is the lack of a translated version of Musicb/Picard a major obstacle to the participation of editors who are not literate in the current languages of Musicb/Picard?
How much more music (Recordings & Events) is added once a translated version is available?

Whilst it would appear prefferable that organic growth and spreading awareness of Musicb/Picard globally would result in user-led translation projects, the pace of this expansion seems very slow outside of Western European languages.
Perhaps deliberate planned expansion into other languages will have great benefits?

Many of the remaing untranslated major languages are spoken in nations where wage costs are ery low compared to Western Europe/USA.
There may be a “low monetary cost for translation” language that needs translating, which has sufficient competent speakers already contributing to MB who could act as liason/quality control/for the project.
It may be that in the music industries of as yet untranslated languages there is both a need and openess for what MB has to offer around Release metadata.
If translation results in Editors joining the MB project, it will also increase efficiency of the Editor group by having native-speakers who can easily find and use native language sources creating and checking entries in that language.

Working out which language to have translated first is an interesting problem.
I’ll leave that to you to consider and discuss.

[Freso]

This is something that I think @Rob should weigh in on (him being the ED so ultimately the one who will make any such decision), but we can discuss it in the community until it gets to the point of a decision having to be made, I guess.

I’m not opposed to the idea per se, but I do see two challenges with it:

1. If we add support for a language for (and thus promote ourselves to) a language community for/from which we don’t currently have a reasonably sized established community, we may end up with an influx of editors who may not properly understand e.g., the style guidelines and who may not share a language with most older/existing editors. This may cause challenges.
2. Paying one-off for a translation may be a good boost and may help out, but we should also plan for a way to keep the translation updated. This could be encouraging new members we may get to help out with translations or it may be planning to periodically (ever X years? every X months?) pay to update translations.

I feel like the 2nd one is a smaller, logistical issue, but the 1st one may cause genuine rifts in the community if different cultures form across different language communities. (But the new language community may also seamlessly blend with the current community, so no. 1 may be a complete non-issue too!)

[Jim_DeLaHunt]

There are major languages in which huge numbers of music recordings exist but Musicb&Picard are not available in those languages.

It seems there is a fundamental question which should get answered first. What is the goal of moving MusicBrainz towards these languages? From the @mmirG’s quoted sentence above, it seems like the goal is extending the content in the MusicBrainz database to include music recordings which are presently not in the database. These recordings seem to have in common that they are in the same few languages.

If more content is the goal, then I would suggest starting by identifying the content we lack, and the languages that content uses, and the markets where that content is used. How many contributors do we have who are interested in that content, and already adding it to the database? Let’s identify those contributors, and ask them what obstacles MB places in front of them, and how they think MB could recruit more contributors like them.

I think we should also spend time thinking about the value we deliver to contributors. What motivates people to volunteer the effort required to contribute good MB content? What people would get value from a MusicBrainz which covers the recordings presently not in the database? What MB feature would deliver that value? How do we find those people and thet them excited about that value?

For me, the value proposition still comes down to tagging music files. I buy a CD from a modern, local classical musician. I want to rip that CD and tag those music files. I add Artist entries for all the people on that CD, and a Release entry for that CD, with the promise that in the end I will have good metadata tags in the music files. (I also like to think building up the content is valuable in and of itself, and that visibility of the artists in the data helps the artists, but that doesn’t drag me to the keyboard.)

So, what is our goal? What languages? What categories of underrepresented recorded music?

[Jim_DeLaHunt]

we may end up with an influx of editors who may not properly understand e.g., the style guidelines and who may not share a language with most older/existing editors. This may cause challenges.

@Freso makes a very good point. I think this points to two extra parts such a project would need:

1. Translating the style guidelines and help docs, in addition to the UI of the MusicBrainz web app and the UI of Picard, and
2. Setting up a mentoring and training system for the new editors we hope to recruit. Language will be an issue here as well. Almost by definition, to potential mentors will not prefer the language in which the new editors will be most comfortable.

[mmirG]

One way of thinking about the expansion of MusicBrainz into new languages is to contrast “organic growth” with “colonisation”.

Colonisation is what this thread proposed - we pick a language and then make it happen by channeling resources that get the work done. The goal is that the new language version will become part of the MB empire and add to our glory.
Historically this approach has had large problems.

Freso pointed out one - we’d need to allocate resources to policing compliance if we want to maintain control.

Jim points out another - the difficulty of educating across cultures and across languages.

The history of colonisation by Western European powers would probably suggest other large problems could arise.

Organic growth is next to be considered. Later.

[IvanDobsky]

There isn’t enough people to keep the English(US) Style Guides and the website in general up to date. It is a small team doing a LOT of work. It would become a huge challenge to just keep all the different translated versions of the Style Guide in sync with each other.

Translations are hit and miss even in the common European languages. There is a surprising number of partially started, never updated work in the translations. I joined the Transifex group to translate Picard to English (British)(Canada)(Aussie). Trivial stuff I know, but it did mean getting to see the translating from the inside.

I can see a situation where someone in the USA points at a style guide, then a French, Dutch and Japanese reader attempts to read the same sections in their native languages. Getting totally different text due to them being out of date. And the more obscure a language gets, the harder the updates will be.

It also makes it harder to update the original USA Style Guide text as any time it is changed it means all the other languages get into an unknown state.

What surprised me the most about the Transifex site is the lack of community in there. No obvious way to talk to other editors. I also expected to get the odd email notification when new text is to be translated.

These are not criticisms - I am just trying to show there is a lot missing even for common languages. Resources and labour are always in short supply. The team already do a lot with very little resources.

[IvanDobsky]

For the OP - currently in Transifex it claims there are 85 Russian translators, 35 Chinese, 0 Bengali, 4 Hindi, 2 Punjabi, 4 Arabic, 59 Portuguese (Brazil), 67 Spanish.

The question is - how many of those are still really part of the project? It claims 605 translators in total, but I get a feeling many of those are long gone. I know there were dozens of names listed under the English sections I worked on, but no sign of anything having been touched for a while.

There is a need for a boot in to the current translations to bring them up to date. Even the common languages could do with some love. Here is the state of translation from the Musicbrainz side.

The trouble is not just getting something translated, but keeping it up to date.

[InvisibleMan78]

The trouble is not just getting something translated, but keeping it up to date.

One of the biggest problems with Transifex IMHO is this:
You can not translate the english/original strings word by word, because you always have to know the context.
Therefore is nearly impossible to do a correct translation without knowing where this text will be visible, how much space you have, what was the action/text before and what comes after this translation.
Then you have to “respect” the work so far. Many of the previous translators used different styles. Which one do you follow?

Just an example:
How would you ever translate this text to your - non english - language:

Both Other and Uncommon are shown in the “other” section, but Uncommon should be used for scripts in Unicode which are unlikely to be used.

[mmirG]

IvanDobsky you are doing brilliant work here.

Getting a functioning translation maintenance system happening for a active comunity of translators is a prerequisite to having more MB translations and maintaining the translations we have.

And it appears from what you write that the Style Guidelines need to fit through a new hoop (metaphor meaning “meet a difficult new criteria” ) - they need to be accessible to efficient translation. Have I got this correct?
reosarevkok will love this one

What else have you discovered around getting maintaining and translating happening?

What priorities do you see around maintaining translations?
Once we’ve got maintenance then we can move on to new translations.

Might it be a fun idea to have A Month On Translation Maintenance as the next Community Project?

[mfmeulenbelt]

One of the troubles with translating and MusicBrainz is that, by nature, MusicBrainz is very collaborative and translating is very hard to do by commission. So the current focus of “getting as many translators involved as possible” may not lead to the best results.

If more than one translator is working on a translation project, the only way to maintain consistency is to a. have clear style guidelines per language and b. to have a single editor check and, if necessary, fix everything.

This may not be totally achievable for a project like MusicBrainz, but it is the ideal situation we should strive for.

On the original topic of paying for a professional translation: it might be an interesting investment opportunity. If we can identify a language that is:

• a. underserved but has a large community of native speakers that is into music and
• b. has a number of companies active in that area who would benefit from being a MusicBrainz customer if MB had a sizeable music community from that area, a translation might pay for itself in the long term.

This would be a multi-year experiment though, but even if it doesn’t deliver new customers, it would deliver a new translation.

[IvanDobsky]

@InvisibleMan78 - totally agree with the observations. Having a plain list of words and phrases sitting totally out of context is really tricky to work with. And the less European a language the more varied that translation will need to be.

As to “other translators styles” - I am always of the belief that work can be improved. So if I see confused styles I would replace them. The edits can always be reverted if someone disagrees.

There needs to be some kind of link from the dumb list of text to where it is found in context. Or someone to ask!

But I am really not the best person to talk here as I can just about order a Beer in German, and maybe count in French. Otherwise I stick to English in its various flavours.

When I dived in to correct “colour” in Picard it meant seeing the reality in the translation. It is so strange workin in there. It is like walking into a big warehouse with no idea if anyone else is even in there. I see a dozen names listed as previous English translators - no idea how or if I should contact them. No one seems to be co-coordinating things. So I just posted comments in the forum and found a an Aussie and a Canadian who helped confirm a few phrases.

It would REALLY help if there was some louder feedback from people who read other languages. Is the German version okay? Or is it full on contradicting styles? That would be a good part of a “Month on Translation Maintenance”

It is interesting looking at the list of partially complete translations above as it also does hint towards the languages of the main MB team.

@mmirG - don’t give me any credit here. I am just a loud mouth trying to stir up some action instead of words. I would love to see how much of an “active community of translators” there really is. Personally I would want to light a small bomb under it all and see who actually wakes up.

I expect there are some “translated” versions of the style guide that are many versions behind. IMHO those pages should be reverted to English when that occurs to stop them being misleading until the translations can be updated.

[reosarevok]

I expect there are some “translated” versions of the style guide that are many versions behind. IMHO those pages should be reverted to English when that occurs to stop them being misleading until the translations can be updated.

For what it’s worth, if a string changes, the translation breaks and we do go back to English until we have a new translation, so we shouldn’t be showing old data. Because code gets released to beta a week before it gets made public in the main server, there’s usually a week or so for active translators to translate any new lines before most people see them, meaning if a specific language is active enough, it might never have any issues with people seeing English in non-beta servers.

Of course, that’s also why we’re not currently translating documentation - we don’t have any good way of doing it yet! But I’d expect once we do, it’d be the same, where if a paragraph has changed that paragraph reverts to English until someone retranslates it.

[IvanDobsky]

For what it’s worth, if a string changes, the translation breaks and we do go back to English until we have a new translation,

How does that show up in Transifex? Is there an obvious flag when a phrase has been obsoleted and now not in use? (sorry, I did it again. Too many tangents in one post. Think we need to take your reply and this reply into another thread. My fault for wanting to stir up reactions better.)

[yvanzo]

This is very obvious as obsolete messages are just removed from translated messages. However, Transifex keeps these in history for reference, so that translators can compare old with new messages and just modify old translation accordingly.