https://www.duolingo.com/davidnucci

News from the incubator?

Does anybody can talk us through what is happening to the incubator at the moment? It seems that the numerous exciting new languages recently added are not making any progress (Percentages are stuck at 0 or 1) Also, I noticed that German for Russian speakers has seemed stuck for more or less one week while it was expected to enter Phase 2 a couple of days ago. I'm so excited for Turkish to be available that I keep checking the incubator everyday. Whatever is happening, props to the various teams working on it and good luck!

4 years ago

26 Comments


https://www.duolingo.com/annika_a
annika_a
  • 22
  • 17
  • 17
  • 16
  • 11
  • 11
  • 5
  • 2
  • 2
  • 2

There have been some messages about technical problems with the new (internal workings of) the Incubator, which have slowed down at least the new English-->Other language courses.

4 years ago

https://www.duolingo.com/Lavinae
Lavinae
Mod
  • 25
  • 25
  • 20
  • 13
  • 12
  • 12
  • 11
  • 11
  • 9
  • 8
  • 7
  • 7
  • 7
  • 3
  • 2

We've indeed been facing technical problems. Luckily, many of these issues seem to be gone now.

One of the technical issues we have is that data has moved because of implementing the new tree both for the reverse course and the original one. Some of the data (in the form of added translations) is missing or inaccessible. We've also had the issue of new translations 'leaking' into the old courses, which was quite problematic, and translations not being saved (very early on).

Features are also still being tweaked. For example, until recently words we translated could not be deleted from the lessons we selected. This is one of the issues that made it difficult to fruitfully construct a course/tree, but which has now been fixed. :)

Instructions are also still being developed and since I've found that it's really easy to mess things up when building the reverse tree, it's better to take it slow and be cautious. I suspect that's what the other teams have also been doing, in addition to figuring out how to teach a new alphabet, fix issues with suffixes, etcetera. Fortunately, these are problems that the Dutch course doesn't have. :)

p.s. For those interested in the progress of the English --> Dutch course, we've posted an update about our progress here.

4 years ago

https://www.duolingo.com/ElOtroMiqui
ElOtroMiqui
  • 14
  • 13
  • 11
  • 8
  • 7
  • 3
  • 3
  • 2
  • 2

I have a question I've always wanted to ask but never found someone to ask it.

How do you guys know what you should teach first in the new language?

4 years ago

https://www.duolingo.com/Lavinae
Lavinae
Mod
  • 25
  • 25
  • 20
  • 13
  • 12
  • 12
  • 11
  • 11
  • 9
  • 8
  • 7
  • 7
  • 7
  • 3
  • 2

I can't really speak for the other teams. :)
However, this has been the Dutch team's approach so far:

We have been studying and will use:

  1. Comparative material - other Duolingo and language courses/material
  2. Professional experience/Scholarly knowledge
  3. Statistics/Feedback

First, we've been looking into the structures and content of the other courses, and that of the English --> German one in particular, while consulting our mentor (who also happens to be one of the experts of that course). We're using the English -- German course because that is, in our opinion, the course which will be most similar to that of English --> Dutch.

I also have access to Dutch course materials through the company I've worked at. I've been going over the content and structure of these Dutch courses for English speakers. In addition, we have access to University sources (from the department of the Neerlandistiek in particular, the department of Dutch studies).

Second, we'll be using personal and professional experience. In my team there's a linguist, a translator, a computer scientist and I have professional experience teaching Dutch. I think that together we'll make the perfect team for building this new course, not excluding the possibility that we'll call in extra help (which we intend to do, when needed).

Third, once the Dutch course is in the beta stage, we will have access to user statistics. We will be able to see where students make the most mistakes and which changes should be made for improving the success rate/quality of the course. We'll be using these statistics and feedback received in the beta course for improving the course further. :)

Of course, building a new course for teaching a new language is a novel experience and, in that sense, a learning process. I could never argue that we know how to build a new course. However, we do have a good idea what to look for and, what I consider to be, a good basis for creating this course. Furthermore, we will have the ability to adjust the course to our learners' needs and make it the best it can be, mostly because of the access we'll have to the success and learning statistics of our users.

4 years ago

https://www.duolingo.com/ElOtroMiqui
ElOtroMiqui
  • 14
  • 13
  • 11
  • 8
  • 7
  • 3
  • 3
  • 2
  • 2

In my team there's a linguist, a translator, a computer scientist and I have professional experience teaching Dutch.

"Wow" is the only thing I can think of when I read that phrase.

I'm glad that Duolingo has teams like yours!

Thanks for taking the time to respond to my question in such a detailed way!

4 years ago

https://www.duolingo.com/Soolrak
Soolrak
  • 25
  • 25
  • 25
  • 24
  • 23
  • 22
  • 20
  • 20
  • 20
  • 19
  • 18
  • 17
  • 15
  • 14
  • 14
  • 12
  • 12
  • 12
  • 11
  • 10
  • 71

Haha. I said exactly the same, plus a ":-O"

4 years ago

https://www.duolingo.com/CuriousAtanaa

Agreed, that sounds like a fantastic team.

4 years ago

https://www.duolingo.com/RymeLegis
RymeLegis
  • 11
  • 8
  • 4
  • 3
  • 2
  • 2
  • 2

Awesome. I'm looking forward to the Dutch course and I appreciate the efforts of you and everyone on your team. Dank je wel!

4 years ago

https://www.duolingo.com/jitengore
jitengore
  • 20
  • 16
  • 16
  • 13
  • 11
  • 8
  • 6
  • 6
  • 5
  • 4
  • 3
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 1760

Very useful stuff, Lavinae. Thanks for sharing!

4 years ago

https://www.duolingo.com/Dessamator
Dessamator
  • 25
  • 14
  • 13
  • 13

In my team there's a linguist, a translator, a computer scientist and I have professional experience teaching Dutch.

Interesting, you guys seem over-qualified :). I think that at least for the other language courses Duolingo found the right formula, and that's probably why there are many strange sentences. They needed to ensure that only the few words being taught were used to teach new concepts and ideas.

I think one good approach to take is probably to use the tree in levels. For example, from the beginning up to the first checkpoint might be the knowledge a 1-6 year old might know. From the second checkpoint to the 3rd, that might be more or less what a 7-14 year old would know, and the last part takes us to a few years before the end of high school.

I would wish you luck in your course creation endeavours, but it seems that you don't need that because you have knowledge and skill on your side. :)

4 years ago

https://www.duolingo.com/Lavinae
Lavinae
Mod
  • 25
  • 25
  • 20
  • 13
  • 12
  • 12
  • 11
  • 11
  • 9
  • 8
  • 7
  • 7
  • 7
  • 3
  • 2

I'll keep that level division in mind! Thank you :)

4 years ago

https://www.duolingo.com/TaniaForsh

Hi! I was wondering when are you expecting to launch the Dutch for English course? As October 2014 (or 2017 for Russian-English) does not sound promising. Thank you for the hard work!

4 years ago

https://www.duolingo.com/Lavinae
Lavinae
Mod
  • 25
  • 25
  • 20
  • 13
  • 12
  • 12
  • 11
  • 11
  • 9
  • 8
  • 7
  • 7
  • 7
  • 3
  • 2

No worries, the dates provided by the incubator are not accurate at all. We have barely started and the incubator's estimated date simply estimates the amount of translations done within x amount of days and extrapolates that number. Hopefully we'll be able to get the Dutch course out of beta within a few months. :)

4 years ago

https://www.duolingo.com/AdamNowek
AdamNowek
  • 24
  • 11
  • 9
  • 3
  • 2

I seriously can't wait to help my Dutch-learning with Duolingo!

4 years ago

https://www.duolingo.com/SongbirdSandra

I'm so excited to hear this news! I cannot wait to learn Dutch. Thank you very much and give my thanks to your whole team. Your work is appreciated so very much. :-)

4 years ago

https://www.duolingo.com/jitengore
jitengore
  • 20
  • 16
  • 16
  • 13
  • 11
  • 8
  • 6
  • 6
  • 5
  • 4
  • 3
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 1760

This week, incubator has been fully back in action for the forward courses. The reverse courses ran into a few glitches, but this should be expected of an innovation as incredible as the Duolingo Incubator. With the six reverse courses - learning those languages from English, this incubator journey has moved to uncharted waters. Contributors can now 'structure' their trees specifically for their languages and I am sure the dev team will soon fix all the issues that might have come up (if they have not yet been fixed).

You might have missed to read a very nice update by SomosTortugas for the German for Russian course.

Cheers!

4 years ago

https://www.duolingo.com/davidnucci

Thanks a lot. It's all good news!

4 years ago

https://www.duolingo.com/andrewmof

German for Russian should be complete by Monday, as far a the team are concerned although it says its at 98% their work is complete and the rest of the work is left with the duolingo team.

There has been some minor progress on the new English courses but expect for them to take a few weeks to get off the ground while teams organise themselves (we also don't know if duolingo has provided all the sounds for these new courses yet)

4 years ago

https://www.duolingo.com/ijoni
ijoni
  • 25
  • 11
  • 11
  • 10
  • 7

I think they should add contributors. That way it would be way faster, have less mistakes and reduce the alpha and beta waiting time.

4 years ago

https://www.duolingo.com/ViArSkoldpaddor

There are several things wrong with this statement.

Your assumptions:

"Waiting time depends on the contributors alone":

  • We have been done since last Tuesday. With some luck, we are going to go life today. Maybe not. Without the last incubator change we would have been done a week earlier, but we might have had to wait until today anyway, because the guy who is responsible for a particular piece of code that we need is on vacation. If we had more contributors, we would have just been waiting for longer.

  • The Russian<->English team in Beta is at 103% stability and has been for a week. They are still in Beta. The process of releasing the courses is not automatic.

  • The release process from Beta depends on the amount of users. We THINK that we are done with the course and it is stable. But the stability of the course is measured by number of error reports per 100 users. Unless a problem is glaring (and we think we have fixed all of those BEFORE release into Beta) we will need several reports for one problem to recognize that yes, we have to change something. Than another week will pass so that we can measure whether our change has produced the effect we wanted. Before a certain amount of users is reached, there aren't enough reports for us to work on, and it doesn't matter how many of us there are.

"More applicants will solve problems":

  • there is a significant learning curve to both using the tools AND the methodology of a team. Everything a new contributor does needs to fit into the existing framework built by the team, otherwise the users will be confused by particular types of translations being accepted some places but not others. If we accept a contributor now, depending on the person he will consume our time for the first week as opposed to producing useful work.

  • More contributors will make more different mistakes, not less. In order for them to make fewer mistakes, everyone has to go over the same sentences and fix things of which he is absolutely sure are wrong. That means duplication of work, ie the time doesn't get shorter. Again, we need to be able to trust people to recognize that we all make mistakes, recognize their own, and not "fight for a translation" like users are doing in immersion, because otherwise we will never get released. For this, the team needs to be pretty tight. A team of 20 people is unlikely to be manageable in this way.

"It is possible to accept more applicants":

  • The original team starts with only 1 invite each. IE our team with 3 people has exactly 1 (one) invite left. We would have to go to the Duo staff and ask for more. Except... and this is a big one:

You assume that it is possible to find more fitting applicants.

  • We can't find enough people who are both qualified and motivated to help. Just writing a shiny application isn't enough. We basically need to spend time watching the people, contacting them, etc.

  • Since it is significantly easier to destroy work than to produce it, we can't take anyone who we can't trust.

A typical example:

  • we have one application that is perfect, but it is obviously written from a fake account, since the application is the only thing that happened from it. The user doesn't respond to requests on his stream. We have no other way to contact him. => There is no way we will take this person as a contributor.

  • We have applications that are written by people who log in once a week. With those, we need to think hard and find out whether they are capable of contributing enough to compensate for the time they will take to introduce them to the course.

  • we have applications that have systematic grammatical errors in one of the languages. Since there is no role separation in the incubator, they will be able to do everything that we can. It would obviously require some work on our part to make sure that in the net they are productive contributors.

In the end, we will have to pick one of them, but it will take work. And we can only hope that we don't make a mistake.

4 years ago

https://www.duolingo.com/CuriousAtanaa

I imangen there are issues with coordinating large numbers of people when doing something as complex as designing a course.

4 years ago

https://www.duolingo.com/davidnucci

Thanks for your answers!

4 years ago

https://www.duolingo.com/tariqnisarahmed

Ditto!

4 years ago

https://www.duolingo.com/duolingo6666666

I heard theirs going to be irish. though i'm not shure when it will come out :(

4 years ago

https://www.duolingo.com/Erven.R
Erven.R
  • 16
  • 7
  • 3
  • 3

There is already, but it doesnt have any contributors

4 years ago
Learn a language in just 5 minutes a day. For free.