1. Forum
  2. >
  3. Topic: Duolingo
  4. >
  5. [WIU] Weekly Incubator Update…


[WIU] Weekly Incubator Update: 2017, Week 25.

Note: Data retrieved on 2017/06/25 at 11:56 (UTC).

I. Phase 1 Courses

I.A. Transfers

  • Exits (Courses reaching BETA Phase): None.
  • Entries (Courses entering the Incubator): None.

I.B. Completion Summary

To enlarge the image: Right-click on it then select "Open in a new tab".

(A) This week's value automatically generated from Duolingo's statistics.
(S) This week's value self-reported by the team of volunteers working on the course.

II. Latest News from Course Volunteers (All Phases)

Course "Hungarian from English" (Phase 2)

By AndrsBrny on 2017/06/23 (full text here)

Big steps closer to graduation!

Hello everyone,

it's been a while since we posted an update. During this silence, we've be...

Course "Korean from English" (Phase 1)

By Soedori on 2017/06/20 (full text here)


"This is not a drill. I repeat, this is not a drill"

We are launching on August 17th. This tim...

III. Previous Weekly Incubator Updates

See this index of WIUs.

June 25, 2017



It looks like our Korean course weekend update came out a little late to make it here, but now we're at 82%


I didn't receive any update from Korean team this week... ;)
Last update I received from a team member was the "70%" of last week, hence it's that one that is used (until either another one or, after 3 week, it'll go back to the automatically computated).

That said, this "82%" is:

  1. the team own evaluation/computation (which, by coincidence/chance, happens to be the same as the one automatically computated by Duo system: https://incubator.duolingo.com/courses/ko/en/status)
  2. or the one automatically computated by Duo system



There's a 2 hours old written update on the page you're linking.


Ahahah, I didn't see (read) it. Thank you!

Well, hence it answers my question: it's not a "team computation" but the automatic computation by Duo systems. ;)


the automatic one just happens to seem accurate at the moment :)


Korean is going to massively deliver. and as discussed in another thread, us web based users won't have to wait for the release of their course, which is wonderful.

do you guys have an estimate on what the 2.0 tree will look like? or is it way too early to tell

EDIT to add: hey duolingo that list of incubator languages is looking small-ish. maybe some additions soon?


I hope the rewrite has kept them busy and that's why the amount of courses has dropped to a relatively low level. The last time we had 21 or fewer courses in phase 1 (which we currently have if Japanese is counted as a phase 2 course) was in February/March 2015. For comparison, the record high is 31 from December 2015. It's not like they are running out of willing contributors and at least the Arabic team has already proved that they are qualified.


It's still a bit early, but I think some people on the team would like to compete for longest course, but that might not be until Tree 3 or later. Tree 2 will depend on how Beta goes, but we'd likely add back in the skills we trimmed to get to an earlier release.


Wow! Great news from the Korean team!


Thank you for this!


Hello to the "Hungarian from English" - responsibles

Do you really believe in the words you published above of making bigger steps to graduation? Maybe it's true, but you are still so far away from the goal of a course, that motivates the people to learn Hungarian. Actually the opposite is true.

The course is still having a lot of quality problems. One of them are those surely funny meant but after a while really annoying sentences about "flying kindergarten teacher"

Az óvónő a magas fa fölött repül.

Az óvónők a város fölött vannak?

A mongol óvónő nem a város fölött repül, hanem a folyó fölött.

Az óvónő a hosszú utca fölött repül.

A ház fölött száz óvónő repül!

Az óvónő a föld fölött repül.

Az óvónő a vízen sétál.

Min repülnek a francia óvónők?

There are still more examples.

And when once the kindergarten teachers were flown out, you took another job group:

Nyolcezer-kettőszáztíz takarítónő repül a város fölött.

A takarítónő az óvónő mellett repül: ez gyönyörű!

Another problem are all those run-on-sentences:

A vendégek felállnak a mellől az asztal mellől, amelyikhez nem jön oda egy pincér sem.

Az előtt az autó előtt nem megyünk át, amelyikben amerikai rendőrök ülnek.

Azok között a városok között áll a nagy hegy, amelyek fölött nem repülnek repülőgépek.

Azok mellől a gyárak mellől költöznek el az emberek, amiben bűnözők dolgoznak.

A sárga kutya nem azok közül a gyerekek közül fut ki az utcára, akik az iskola előtt beszélgetnek.

Ezek a bogarak nem azok alól az ágyak alól jönnek ki, amelyikekben ti fekszetek, hanem a mögül a szekrény mögül, amelyikben sok ruha van.

Azok mögött a hegyek mögött folyik a széles folyó, amelyikeken sok étterem és kávézó van.

Sok különböző helyen dolgoznak takarítónők: mozikban, iskolákban, egyetemeken és börtönökben is.

I do not have to tell you the fact, that longer sentences make the number of possible correct versions of it in Hungarian extraordinary big. The chance of giving a "correct" answer is very small, because actually correct versions are rejected because they are not entered in your database.

And how about the "feltételes mód" or in english: subjunctive. It does not exist in your course at all. But it is very common in Hungarian to say:

"Szeretnék..." - I would like...

If my words sound frustrating, then you got it right. For months I made reports over and over again because of "too long sentences", but nothing happened in this matter. Accept one thing: Duolingo doesn't allow anymore to write comments on a quality report.

So maybe my words here might have an effect... Or maybe not.


Hello to the "Hungarian from English" - responsibles

If you want to "contact" them the best way is probably to write on the "Hungarian from English" forum: it's not obvious at all that anyone from the team is following this WIU. It may be, but can't be guarantee and I'd say there is more chance they read the forum of the course they contribute to.

Do you really believe in the words you published above of making bigger steps to graduation?

I can confirm they have never been that close to fulfill Duo's condition(*) to graduate from beta. So I guess they're right in believing they've made progress towards graduating from beta.
Now, if graduating from beta the end/a good metric? It's another question but their statement is correct.

(*) which they explain in their status update.


Can the contributors even do anything to the run-on sentences or add the subjunctive at this point? Correct me if I'm wrong but adding skills for subjunctive requires a new version of the tree and you have to graduate from beta to start that. Maybe the contributors can disable the run-on sentences but if there are a lot of those, that's going to reduce the amount of words (and possibly grammar) taught.


Can the contributors even do anything to the run-on sentences or add the subjunctive at this point?

Sorry, not sure what "run-on sentence" means (not a native English speaker). If they are the "too long" sentences: yes they can add/remove sentences.

About subjunctive, you've explained the situation correctly! ;)


Thank you for your answer above.

Exactly that is it, what I meant.

And I reported for months "too long sentences" over and over again, and nothing happened - accept the case, that adding a comment on a report like "too long sentence" is not possible anymore since about two months.


Anyone know the current percentages for Haitian Creole, High Valyrian and Yiddish? They are not listed here https://www.duolingo.com/courses/all and for some reason I can't do the right hind click thing on their individual incubator pages to get the percentage completion other people seem able to do. Posting the results would be greatly appreciated.


The chart above does not satisfy you?


No, the stats given in the chart above are in some cases quite different to those officially reported here https://www.duolingo.com/courses/all and I don't know exactly where the figures displayed in the chart above come from.


Deviating from those not necessarily significant official digits is a good thing. We talked about it five updates ago. Of course, you can always just enjoy the official sausage and not care about what's ground up in there.

Our calculation does not just copy the official course completion percentage, and this is why some of our status figures deviate from the previous WIUs, in some cases significantly.

We enforce the minimum "total words" target of 1500 to reduce the impact of partial trees on the calculation.

We only use the completion of localization strings for courses that are the first to teach from the given language, and even then only at one-sixth the weight of the word completion progress.

We ignore the completion of images.


I don't know exactly where the figures displayed in the chart above come from.

From the number of course's lexemes completed by the team. See detailed description in the first WIU with this (new) charts.


It's less convenient than hovering or right clicking, but it is possible to highlight/select the progress bar and text underneath (up to where the Estimated Completion Date is) on their individual incubator pages and copy and paste it into a document.

  • Haitian Creole 6.418870621872767%
  • High Valyrian 27.47364841890513%
  • Yiddish 13.998309382924766%

(These have been edited for brevity from what gets copied and pasted, although I have preserved the full numerics.)

If you can look at the page source, this should show it too.


(These have been edited for brevity from what gets copied and pasted, although I have preserved the full numerics.)

A truly meaningless level of precision for a system that is frequently off by >10%!


Thanks very much. Thats exactly the info I was after

Learn a language in just 5 minutes a day. For free.