Weekly Incubator Update: Tracking Progress from October 18th to October 25th
Six courses in the 90s
Several courses have progressed at a fast pace and now we have six courses in the 90s. It is likely that the Hungarian team's too has progressed beyond the 80% mark, but since they haven't self-reported their number, I have continued to show them at 80%.
Thought for the week
Which course gets to beta next?
PHASE 1 Progress: Total 28 courses
Russian for English - 99% | 99% | 99% | 99% (+0)
German for Turkish - 94% | 94% | 95% | 97% (+2)
Polish for English - 91% | 91% | 93% | 96% (+3)
English for Thai - 90% | 90% | 91% | 91% (+0)
Spanish for Russian - 82% | 84% | 87% | 90% (+3)
Catalan for Spanish - 74% | 82% | 85% | 90% (+5)
Hungarian for English - 80% | 80% | 80% | 80% (+0) ^
Vietnamese for English - 64% | 66% | 67% | 74% (+7)
German for Italian - 72% | 72% | 73% | 73% (+0)
Spanish for Chinese - 44% | 57% | 63% | 71% (+8) *
Spanish for Italian - 64% | 64% | 65% | 65% (+0)
Hebrew for English - 59% | 62% | 64% | 65% (+1)
Swedish for Russian - 54% | 55% | 56% | 57% (+1)
French for Arabic - (New) 40% | 44% (+4)
Romanian for English - 40% | 40% | 40% | 40% (+0)
German for Arabic - 37% | 38% | 39% | 39% (+0)
Swedish for Arabic - (New) 37% | 37% (+0)
Greek for English - 34% | 34% | 34% | 34% (+0)
Italian for Portuguese - 26% | 26% | 26% | 27% (+1)
Hindi for English - 21% | 21% | 21% | 21% (+0) ^
Klingon for English - 19% | 19% | 19% | 20% (+1)
Portuguese for French - 18% | 18% | 18% | 19% (+1)
Czech for English - 10% | 10% | 11% | 11% (+0) ^
Yiddish for English - 8% | 8% | 8% | 8% (+0)
Welsh for English - (New) 1% | 7% | 7% (+0) ^
French for Chinese - 5% | 5% | 6% | 7% (+1)
Guarani for Spanish - (New) 0% | 3% | 7% (+4)
Esperanto for Spanish - (New) 0% | 0% (+0)
Mean - 0.88% | 1.15% | 1.45% | 1.5% (+0.05)
Median - 0% | 0% | 1% | 1% (+0)
* This week's Leader Extraordinaire!
^ The Hungarian, Hindi, Czech, & Welsh teams' progress is as per their own calculation
Here's what the contributing teams have said during the last week:
This extra update addresses the growing issue of whether this course needs additional contributors or not. Yes, it would benefit from additional horsepower. No, it definitely is not a matter of just throwing more bodies at it.
First, the would-be contributor’s command of both languages must be solid. This is about the only area into which the applications actually provide any insight. Simply put, if there are errors on one or both sides of the application, the probability that the applicant will work out in the end is low. (An exception in this case would be a strong native English speaker who has somehow managed to learn passable Czech.) So let’s assume that both sides of the application are without major flaws. What else can anyone possibly be looking for? A lot.
The candidate should also already be familiar with Duolingo. Then the rest of the team will know that they will not have to spend a lot of time explaining the basics to their new colleague, and the candidate will not realize too late that Duolingo is just not the right thing after all. But let’s expand this. Ideally the candidate will also have demonstrated staying power. We do not need someone who rushes into things this way and that but does not see anything through. Both the familiarity with Duo and the staying power will be estimated from the applicant’s user history on Duo, not from the application. An application from someone who has spent a whole day here will be viewed as more risky than one from a user who has shown months of activity. A long streak is helpful but not required. We realize that having a long Duo history likely means an interest in a third language or two.
Another thing that may show that the candidate is the right fit is a history of consistently insightful comments in the sentence discussions of the English course for Czech speakers. (When I think about this entire post, I would say that the discussion comments could demonstrate almost everything we are looking for, including the ability to actually cooperate with us.) In other Duo courses, the more informed users often help those who are struggling, and the course contributors have mostly cleared out of the discussions. Many of those helpful users have been invited to the Incubator. In our English course, the Incubator contributors still do most of the answering, even for the most basic questions. From time to time, a question actually goes very deep. Does this not seem a like a wasted opportunity for the candidate whose application speaks of qualifications in both languages and a strong interest in helping others? (Little time to help? We get on that thorny subject next.) The fresh sentence discussions can always be found here.
Finally, none of the qualifications and user history matter at all if the candidate actually does not have, or no longer has, enough time to contribute meaningfully. The ugly truth is that, for this course at this stage, two hours per week in the Incubator will not make any difference, but two hours per day might. Why is this?
Let’s have a look at the some screenshots from the Incubator. The first one shows the configuration of our tree as of today:
This part of the tree teaches some 420 words in 15 skills. Many courses on Duo have about 2,000 words in total. I have been using a 3,700 word goal in the completion estimates. In order for the Czech word to count as “completed”, it must have at least three Czech sentences associated with it. To see what that means, let’s look inside the “Colors” skill next:
Here we see that the various forms of the Czech word for “red” (shown selected in the left portion of the image) currently have four Czech sentences defined for them in the center column. These forms count as one word and only differ by their endings, which reflect the gender, number, and case. The sentence selected in the center column, “Moje oblíbená barva je červená.”, is translated to four different English sentences in the top right part of the image. Now, if we worked with a language more comparable to English, this would not be so bad. But look in the bottom right part of the image, where the main English translation needed to be back-translated to Czech. There are 204 accepted unique translations shown there (when the compact notation is expanded). Some sentences may have only a handful of back-translations to Czech, others a thousand or more. (As an inflected language, Czech possesses quite an insane degree of word-order flexibility.)
Let’s use 100 back-translations as the average. For 2,000 words with three Czech sentences each, each of those with 100 back-translations, we get 600,000 Czech sentences to enter by hand, maybe with some cut-and-pasting. Can we take a mere 30 seconds for each of those Czech sentences, on the average, to think through and type/paste/edit? (This may be light.) I get 5,000 person-hours. Let’s be optimistic and take 5 contributors all active at two hours per day, day in and day out, as I dared to dream above. How long before these heavy hitters get the course to beta? I get 500 days. And this has no allowance for research, planning, quality control, revisions, etc.
So that’s why.
We just cannot afford to waste time if the applicants are unlikely to work out, and so we are holding out for a lot while admittedly offering little. Please help us choose wisely. All the clues are in this lengthy post.
And maybe we will be seeing you in the Incubator!
Last checkpoint passed!
It has been a while since our last update, but this milestone has to have a special update: WE'VE PASSED THE LAST CHECKPOINT!
A few more details about our progress:
We've reached 64%.
We already have 1,371 words completed.
And a special bonus: A new contributor is about to join us VERY soon!
Also, we would like to thank RotemK for contributing to this course. Thank you, Toda!
You are more than welcome to join our Facebook group here.
So stay tuned! :D
Tree 2.0 and Esperanto for Spanish speakers
Biggest news first, Duolingo granted us access to our Tree 2.0! That means we can finally optimize the order of skills and which words are taught when. We look forward to making our course even easier for new learners, especially the first skills.
In other big news, we're grateful that Duolingo has initialized the next Esperanto course; this time for Spanish speakers! If you have any Spanish-speaking friends who want to learn Esperanto, they can sign up to be notified when the course launches. Please be patient as we sort through all the candidates to find the best people to create the course.
Last week, Junesun and I had the joy of visiting Duolingo HQ in Pittsburgh, PA and meeting with the great people there! We'll post some pictures soon to the Duolingo Esperanto forum. :)
Back on track
This last summer we have been through a few things: traveling, moving to new places, etc.. that slowed us down. However, now we have new additions to our team and we are back on track. We plan to release the course in Beta at some point during next Spring. So, we are not really close but we are moving forward! Of course we'll keep you updated.
A presto e grazie per la vostra pazienza!
Week 20: 21% complete, beta launch 31-Dec-2015
We've added over 1,000 sentences this week, which gets us up to about 15% completed. We're still having problems with adding images but this seems to be a general issue.
We've had some significant interest from the Welsh language media this week. We had a spot on the late afternoon news programme and we had articles about the Duolingo Welsh course on the website of the main Welsh language news magazine and the BBC Welsh language website.
Here are the links, google translate does a pretty good job with the articles. The Radio spot ( starts at 57:15 listen for the word Duolingo as the item is presented).
The news magazine.
The BBC page.
6 courses in the 90s. The median continues to be 'one' and the average progress is at healthy 1.5%.
The next update is expected on Sunday, 1st of November at 3:00 pm UTC.
Previous Update 11-Oct to 18-Oct
Wow, Polish had bit of a jump this week! I was expecting it to take much longer for the rest of the progress, now that the school year has started.
I hope to see the Polish and Russian courses released soon!
Edit: If the 96% for Polish is correct, that is.
Since no one has commented yet, Very Impressive team Vietnamese! You are making consistent progress! I'm really looking forward to learning a tonal language :)
Wait, Vietnamese says 74% here but on the course selection screen it still shows 67%. You shouldn't get your hopes up too much. A 7% increase in one week seems too good to be true and probably is. Maybe they are adding and removing skills.
I see that! I also see that there are 8 course contributors now, so I'll continue with the optimism :)
The incubator page however now shows 76% - https://incubator.duolingo.com/courses/vi/en/status
They also have an update not included here.
I look forward to that too. I hope to learn vocabulary and easy sentences. Gramar rules seem to be very easy for what i have seen so far.
That was a great insight into the Incubator from the Czech team, though I suppose they have it a little tougher than some other languages.
Flexible word order is definitely a problem for the Duo system. Such fun to get a peek into the course creation tools, though!
OMG Catalan and Polish ! And 6 courses over 90 thats really great to see ! Any news from the Hungarian team ? Did they give up or something because in the course page it is 99% (even if it is 80%) at least it means that they are working on it. And Go French for Arabic
Catalan and Polish are meant to stay together :). Spaniards often refer to us Catalan people as "polacos" (Polish), for we speak a different language and tend to be considered (by ourselves or by them) as foreigners in their own country.
Translation of the ES-CN update:
The course is being completed quickly!!!
I'm very happy to announce that after the fourth week of hard work, we have already completed 60% of the course. At this rate, the Spanish course will [be born/go into beta] soon. You can become a member of our Facebook group.
Thanks for the update! But where did you get the calculation for Polish? For me it still says that it's at 93%, and they didn't post anything on their incubator page.
I use a Tapermonkey script to read the progress. But one can also do an "Inspect element" at the progress bar.
Oh I see. Also the Welsh team just posted an update like 10 minutes ago, their about 15% percent done they say.
Yes, I saw that shortly after I posted this comment. Let's hope we'll see some progress soon.
Woah, I didn't know it takes so long to get 1 word added to the Czech course. I guess I'm gonna have to be more patient.
And just think...the further the language gets from English, the more possible alternatives exist normally. The Turkish course maxed out with over 3,000 translations on a few sentences :)
I appreciate the insightful update that Team Czech did. It was really nice to get a better look behind the scenes.
Even though I'm going to be occupied for a long time on my current languages, I'm super excited to see all these courses nearing completion! The race is on!
Wheeee, Esperanto 2.0! My first opportunity to earn back a golden owl after an update. ;-)
Yes! Guarani progress! I can't wait :3
FYI, the Welsh team just added a new update. https://incubator.duolingo.com/courses/cy/en/status
Thanks for letting me know. I will add their post to the compilation above, but will not be able to update their percentage. That will only show up on next week's WIU :(
Thanks for adding our update, next week I'll make sure it's there much earlier :-)
Hi jitengore, thanks for another update. When you added the Welsh update your own tl;dr section got caught up in the end of the quoted Welsh update. Probably just need another new line to separate it out. It's only a really small thing and still perfectly readable so not really necessary to fix but just thought I'd let you know in case you want to.
But next week you can still list "15%" for this week when you make that graph, right? ;)
Vietnamese for English 7% increase! It's sooo close I can almost taste the Vietnamese...
Can't wait for Russian, Polish, and Hungarian! Also, it's great to hear that Romanian is back in action!
Where does it say anything about Romanian? It didn't make progress this week and I didn't see an update.
Before Russian got their TTS in, I would have said that Guarani for Spanish speakers was going to come out before Russian :-P
But now that their TTS is in, I think we'll have Russian soon.
Truth be told, I knew a few weeks ago that the Text-to-Speech API was integrated, paid for and was going to go live soon. However, since I had little information on the soonness of that "soon", I preferred to wait for the results before making any announcements (even the team did not know much).
I think no one really said "soon". It is just that when everything's done, you expect it to work rather soon — maybe within next week or maybe 2-3 weeks, but certainly not by the time the thaw sets in.
The core problem was that the Russian company I found did suprisingly little to be ready to work with western customers. They do not even have an English site (I only hope their manager understands English fairly well). I do not think Duolingo expected encountering a company that actively does not want to take your money without a fight :)
They do not need an English website, because you can just go learn Russian on duolingo. :-)
I'm very glad, Duolingo accepted the preparation of the course "Esperanto for Spanish speakers". Thank you so much, Duolingo!
And I am still happy the course "Esperanto for English speakers", https://www.duolingo.com/course/eo/en/Learn-Esperanto-Online already got more than 160k Learners. :-D
Just a minor observation: The TL;DR section is mixed with the Welsh post... =]
Thank you for the post again, Jitengore :D