Grants talk:PEG/Kruusamägi/Minority Translate

Can you provide at least one example of what you intend to do? As I understand it, this is simply the writing of templates that will be pasted into the top of each newly translated article in a group of thematically rated articles (individually tweaked?). I query the benefit of this, and why you don't just shift data into Wikidata if encyclopedia article style is too slow in some instances. Tony (talk) 02:34, 8 February 2014 (UTC)

Templates? Wikidata? What? Kruusamägi (talk) 15:48, 8 February 2014 (UTC)

I can't get the remotest handle on what this would mean, what the outputs would be. I'd like an example or two. Tony (talk) 02:01, 9 February 2014 (UTC)

It would be a standalone program. Just double-click, select the languages you want to use (and category for articles, if you prefer) and you can start. You have one window within there is everything necessary to make the translations -- just write the lines, press "save" when you are done with one article and another emerges. Program will also deal with Wikidata and do the additional "paperwork" so that used can only concentrate to the content. You could use several Wikipedias for the basis of translation, program can make suggestions, etc.

For example you want to translate from Estonian Wikipedia to Võro Wikipedia. You would like to take only one topic per day and so you select for instance the category 'Võrumaa kirikud' (churches in Võrumaa). One-by-one you translate all of thous articles (or with just one click you could decide to skip some).

As it is a separate thing, then it will be able to remember all of the user preferences and configurations. It can even be used when there in no internet connection at a moment (but it would later still need a connection to send the data). Kruusamägi (talk) 18:21, 9 February 2014 (UTC)

Questions

Thanks for submitting this proposal. I have a few questions related to this project.

• Do you have any working prototype?
• Does this tool can translate automatically?
• Did you tried to use the Translate Extension (https://www.mediawiki.org/wiki/Extension:Translate)? what are the difference between these two tools?--Nasir Khan Saikat (talk) 07:27, 8 February 2014 (UTC)

• No prototype at a moment. A working prototype would mean that the thing is practically ready and then there would not be a need for a grant.
• It is for helping the user at manual translation. It should be able to provide some help (for instance when there is a dictionary available), but not translate the whole thing (not do mention, that this isn't possible at all for small languages, specially with the ones that have a complex syntax). Bot articles don't help small language versions and what they need, is a real content. This could only be provided by people who know what they are doing and the more help we can provide for them, the better it is.
• Translate Extension in mainly for software translations, but this tool will be designed for articles. Kruusamägi (talk) 15:48, 8 February 2014 (UTC)

Thanks for the reply. Translate Extension is not only used to translate the software it is used to translate to articles. Where the TranslateWiki is used for software translation. So here my suggestion is before staring to build a new application from scratch you should study the existing ones and talk with the developers who are working on it. --Nasir Khan Saikat (talk) 09:38, 9 February 2014 (UTC)

Main thing is that this 'tõlkevidin' would be far easier to use. Really, this translate extension is so complicated, that most of the people will never be interested to learn to use it. It still is a wonderful thing and you can do many things with it, but that is also the reason I don't see no point to redesign it. So, there are lot of things you can do with it, but is isn't perfect for translating articles. There needs to be something that is designed for articles and that is simple enough that my grandma could use it. This is what I am aiming here.

After this idea emerged, then I talked with Niklas. That was the reason I didn't started before, as I did understand that there is something coming right away. This was nearly a year ago. Not to mention, that what is being developed, is not the same thing anyway. So if one day there might be a WMF tool for translating articles, then it would not be a competitor of this currently proposed program, but more as a additional possibility, that would be used in different situation (within bigger Wikipedias and with longer articles). And it is a web-based thing. Kruusamägi (talk) 18:21, 9 February 2014 (UTC)

WMF Comment

Thank you for your engagement on this grant so far. I have connected Ivo with the WMF Language Engineering team to discuss the viability of this idea. Looking forward to hearing the results of their discussion. Alex Wang (WMF) (talk) 02:00, 12 February 2014 (UTC)

Measures of success

Let's link the WMF project in question please: mw:Content translation. :) Some research and design information was added to it, it's important that we all stay up to date with each other. In fact, I find the following argument very compelling: «It's a mistake that we didn't proceed with the project back then as we could had created this within 1-2 months and it would have been ready by the middle of 2013. We'd like to start the project now and not wait any longer». However, I would like to propose a couple tweaks:

• in the premises, highlight what's IMHO the biggest strength of your project i.e. that it has a community/language supporting and ready to go with it and that it will be developed as a JavaScript gadget (right?) so with very flexible and fast development;
• as a measure of success, I'd add that the tool be used by at least X users, with at least Y of them submitting feedback, and that a report is produced with the learnings from the project on what can make such a tool effective, within 3? months from now;
• secondarily, if Apertium pair(s) for the languages of interest here (fiu-vro to fi and fiu-vro to ee I suppose? also fiu-vro to English probably) are not available, that would be a goal too.

As a community member interested in translation, my reaction given the modest cost (compared to several IEG grants for wiki editing and the like) is "let's give it a try!", but there is already a plethora of such tools for translating Wikipedia articles, so what would really be valuable is the design input that such an experiment can provide for the tool being developed by the WMF, which is more scalable and hence taking much longer. For that purpose it would however need to come before the content translation extension receives a lot of development work. Finally, I hope you will eventually contribute to MediaWiki itself in related areas, there is a great need of volunteers. --Nemo 10:10, 14 February 2014 (UTC)

I was already wondering why no-one hasn't added that link :)

Some of the grants have surprised me as well. For instance 'Wikipedians in European Parliament' (but as I got the images of Estonian Parlament for free, than I might be a bit biased here).

But now some answers:

• It would be a java program. The goal is to make the use of it as easy as possible and maybe this simplicity could be the biggest strength. But it really is hard to tell, what would be the most import, as it is nearly impossible to project how much it will be used (and where). Maybe the biggest strength will be that it involves linguists and we hope that it would also be used as a tool by them -- that would get a lot of new people interested in Wikipedia. One idea is that this tool will enable to mark down exact translations and therefor it will create a parallel corpora for languages that have barely nothing like that so far. Estonian linguists will definitely find it useful. (We would store thous "this sentence equals this sentence" in our server and make it freely available.)
• Producing "learnings from the project" might be possible within 4 months from now (from Estonia). I also hope that I could successfully promote it to other Finno-Ugric people rather fast, but getting feedback from them would take more time.
• Estonian language is far more complex than English, that makes a machine translations very challenging. And with Võru (or Võro) it is even more difficult, as there just isn't much tech support behind a language what is spoken by 70 000 people. But as I know, then there is something being developed in the University of Tartu and it already rather good.

In Estonian Wikipedia translations are actually rather rare. It is normal that some content is also taken from other language versions, but it is by far more common that the articles are made from scratch. So I don't believe that full article translations will ever become very popular. But in small Wikipedias most of the articles are very short anyway and it is somewhat more important to first get the quantity up and build the structure of articles for future. Fastest way to do this is by using some other Wikipedia as source and translating most important sentences about a topic. For instance Võru is rather similar to Estonian and almost all the native speakers speak perfect Estonian as well (as the last is pretty much the only language that is used in schools). Some people consider Võru a dialect of Estonian, but Southern-Estonian dialects (like Võru) are quit different from written language (that is based on Northern Estonian dialects), and so they are slowly starting to get recognition as separate entities). So if there is a simple tool, then it might be possible to get thous people contributing, that have never before edited Wikipedia and grow the user base. The tool won't be very useful for the development of Estonian Wikipedia, but Võro Wikipedia only has 2 (very) active users. So even 2 more would have a significant impact. Kruusamägi (talk) 22:59, 14 February 2014 (UTC)

Ok, good points @Nemo, I was involved as a linguist, and I played a passive role in the grant, so I am not able to comment on all the differences between the apps. I fully acknowledge the work on ContentTranslation, and I've also used it with great experience. Perhaps in the future also ideas from Minority Translate could just be combined into ContentTranslation. But I also see a benefit of having an offline application that is only loosely connected to Wikipedia. Doing translation work shows me great potential to include people who normally don't write to Wikipedia, especially among speakers of smaller/minority languages. I believe that this deserves a platform of its own for campaigning and technical developments. Mobilizing native speakers in small communities could bring many benefits to language vitality and status in digital and online environments, and I think both the communities and Wikipedia would benefit from intentional collaboration with linguists and language activists around the globe. We've tried to start with the small initiative, but this is definitely only an early step, with real progress only possible if many people get involved on both the side of development and gathering of knowledge, as well as usage and local distribution and demonstration. Our effort is still making only basic strides, but I believe that this effort could be worthwhile particularly in mobilizing small and sparse communities in writing to online communities in general and Wikipedia in particular. Contributions to Wikipedia is also a goal that they might be able to personally relate to. The main aim of the project for me is to "sell" the activity of writing into Wikipedia in these circumstances. This could be done by whichever app, but for now we have a possible start in Minority Translate. We may be looking forward to some results of active campaigning in the next 3-6 months. An interesting data point to consider is how many relatively new users could be seen using this, or whether their productivity could increase. Comparisons with ContentTranslation are also very relevant here, but I wouldn't bury the idea of having many different applications to support many different work-flows that the users may want to have. I'd be happy to discuss these questions further too. Thanks for your input. Pusle8 (talk) 11:49, 27 February 2016 (UTC)

Notes upon approval

This request is approved. We are supporting this grant because it is an innovate experiment in addressing language barriers and will hopefully encourage small language communities to contribute to Wikipedia. However, we do think it would have made more sense to create a web application and hope that the project is so successful that one day it will be turned into a web app (or at least inform the development of the current WMF translation project). We also do not expect the tool to work with all languages, especially non-Latin, non-right-left languages. Please note that care would need to be taken if and when the tool is expanded to include other languages. Alex Wang (WMF) (talk) 04:44, 26 February 2014 (UTC)

Request for extension

So that we could use the funds prior to the end date of the grant, we'll like to postpone the completion date. 2 weeks should be enough and therefor we would like the new completion date to be 15th of April. Reason for extension is that the paperwork has taken longer than anticipated and it isn't likely that we'll receive the funds before the current end date -- that would make the use of thous funds impossible before the 1th of April. Project itself is developing almost in the speed that was planned for it and there hasn't been any real problems with it so far. Core of the program and first testing should be ready by 31th of March as planned. Future development, testing and promotion of the program will be done in the coming months and they aren't involved with the current grant request anyway. Kruusamägi (talk) 23:03, 27 March 2014 (UTC)

This request for extension of the grant end date until 15 April 2014 is approved. Alex Wang (WMF) (talk) 23:09, 27 March 2014 (UTC)