[#DuolingoForumGems originally posted on 2014-07-15 on the Duolingo in English forum by AureliaUK
This post is interesting as a slice of Duolingo history, obviously, the situation changed since then :) ]
Hungarian for English - suddenly at 76% ???
Is this real or a glitch in the system? Just to clarify, it was showing 20.7% as at midnight.
http://incubator.duolingo.com/courses/hu/en/status
comments:
They deleted a lot of the standard words/skills. Therefore the percentage is totally incorrect, a realistic value is probably still somewhere around 20%.
- 25
- 25
- 25
- 22
- 15
- 13
- 12
- 11
- 8
- 6
- 3
- 2
- 2
OK, thanks. I wondered if it was a gremlin caused by the crash earlier today.
VÁLASZBEJELENTÉS.TÖRLÉSLINGOT ÁTADÁSA•7 ÉVE
comment by blazmah:
We saw this coming, haha.
Just like sommerlied said, we deleted all the standard skills, which decreased the total number of words in our course. The percentage, currently, is calculated by dividing the number of completed words with all words, thus the 76%.
You ask why? Well, we actually asked ourselves: why not? We are building the course from scratch, and never ever used any of the built-in standard skills.
Furthermore, while the original courses have around 1800-2000 words we don't know exactly how many words we will have, and expect about 4000 words. Thus we've had few options:
1) We gradually delete standard skills (that's what we've been doing until this very moment; it would have caused constant growth of our course, but in the end we would move around 85-95% for weeks)
2) Delete all built-in skills. (this would cause a huge increase in percentage, then it would cause stagnation)
3) Add all words without deleting any skills. (this would cause a huge drop in percentage, and continuous stagnation)
We chose the second option. The way Duolingo calculates percentages is clearly a bad one, our act is the perfect way to prove that. For that we wouldn't like to rely on the current system as it is clearly faulty.
Where are we then? About 11-12 percent of our course has been done. From now on forward, the best option is to self-report the progress. We follow a professional book and build our course basing it on that book and its content. Thus our course will be extremely different and much longer than the "usual" Duolingo ones.
As for what solutions we recommend instead of the current system: well 1) we could report our progress ourselves weekly, not letting the computer to do so, or 2) set up a "goal" number of words instead of an ever changing one, and calculate percentages to that one.
- 25
- 14
- 13
- 13
I think the best way is to use project management guidelines. Set up milestones and report them (at least internally) as you reach them.
Also you should beware there are two ways you can proceed, focus on your progress (like the Ukrainian team) with little user communication, or communicate with users and make changes or adapt the course content based on user input (like the Hindi team, Dutch, Danish, and Irish teams seem to have done).
The disadvantage of communicating too much is probably that you raise the users' expectations, and they start "communicating" their expectations too many times. But anyway, you guys have the benefit of being incubator veterans, so you've probably worked out your incubation process already.
- 18
- 2
What book are you basing it off of? It might be useful to pick up the book and start reviewing before the course opens.
I don’ think it’s a secret, the book we are basing the basic order of grammatical structures we teach on is “Erdős, Kozma, Prileszky, Uhrman - Hungarian in Words and Pictures”. Must point out, however, that the vocabulary we’re teaching is rather different, as - while this is an excellent and well-built book - its vocabulary often reflects the communist era it was written in, which is of course not something we want to do.
Also, there is a huge difference between learning from a book and from duolingo, so while the grammar is introduced rougly in the same order and structure, the material had to be adopted to Duolingo’s mostly “theme-based” approach, not to mention its limitations when it comes to words/grammar used in same or different levels of the tree (for instance, you can’t use a word if that was introduced on the same level but in a different skill).
In other words, while the book - any book - is teaching the language in a highly lineral way, direct adaptation would mean having one skill on each level - which is NOT how any tree in Duolingo does or should look. Maintaining this flexibility and applying a workable logic while following that proven structure of grammatical introduction is not an easy task...
So... if you do look at the book, look at it as a source, a base, or even a reference, and NOT as “the” material you’ll learn here...
I think you should post this on the Incubator page, as a lot of people are confused/excited by this apparent progress.
Good idea, thank you! Just wrote the first one. :) By the way, from this day forward we will write updates weekly in which we will self report our progress, that way people can still excitedly watch our progress even without Duolingo's automatic numbers. These updates are due every Sunday.
I have to ask.. When do you think it will be done? I know it's hard to determine an exact date but could you maybe guess a month?
No, and I do not even wish to do so.
Because it won't be done for a while?
You got it. Implementing roughly 4,000 words (you guys are crazy!) will take a lot of time. They will be really fast if they will have it done by the end of 2014.
The “4,000 words” might be a bit misleading. Hungarian grammar is relaying heavily on conjugation. A verb has over 70 forms, and there are countless conjugations you can do on nouns and adjectives. Although we can group them in certain ways, as Duolingo works now, we must add the same “base word” more than once - even several times - with different conjugations. That means that our estimation for the “base words” used is about the same as an English course - say 1900-2000 words, but with all those conjugated versions, the “word count” could go up to 4,000 or even 5,000.
As Nitram said, we’re following a professional book that has 25 lessons and we’re working on lesson 3 at the moment. The language is very complex and we must build it completely from scratch, can’t simply rely on the given English template. While it might sound like say 10% is very low, it used to sit at 0% for a long time because the Incubator was simply not capable of handling the needs of building a course from scratch, and now that it does, we’re progressing steadily. We do hope that it won’t last until the end of 2014, we hope that we can come out to Beta much earlier, but saying any real deadlines now would be very difficult.
You're right Alex, there are a lot of challenges in front of us that we probably can't even see yet. Having said that, we're progressing steadily, at a hopefully increasing speed, and so we are optimistic. Thank you for your support!
@boringtomi: I'm aware that Hungarian is a pretty tough language to learn. I once studied Ancient Greek some time, thus I know what a complex language is. There we had countless verb forms as well.
It's awesome how you progress, though and I also hope that you will be much faster than December, but realistically speaking, we often tend to wish to get things done, but I suspect that there will be further difficulties that could hold you back, be it server issues, glitches, incubator problems a.s.o. - but I'm pretty sure that you guys will build an awesome course!
No comments:
Post a Comment