How "smart" is Duolingo at grading answers?
Just out of interest, how smart is Duolingo at assessing your answers? One of the great things about this course is that as it progresses, the sentences get longer, really testing you in the language. However, I would imaging that means that there may be quite a lot of different valid translations for a particular answer.
Is it up to the course creators to literally put in every possible response, or is Duolingo able to correctly assess slightly different sentences?
Also, sometimes I've noticed that the website takes quite a while to mark and answer as right or wrong. Now, that could be simply due to the load on duolingo.com or perhaps my internet connection (although I do have a good connection), but I was wondering if there was some sort of "manual verification" taking place, and that's why it was taking so long??
it is up to the course creators to literally put in every possible response
sometimes it takes long when there are too many alternatives (we can add up to 3000 at the moment)
The problem is that unlike other languages Turkish is very flexible in terms of word order because of its agglunative nature. I guess this is essentially what makes harder to provide every possible correct answer. (You can almost calculate it for each sentence with permutation rule) It is understable that Duolingo cannot handle it automatically now because this is not a big problem for most of the languages and it's also hard to construct an algorithm for it. A perfect solution would be impossible anyway. So everyone should keep in mind that the course uses the neutral word order which is preferred in written language instead of providing all the possibilities in daily speaking language. And if you look over the discussion topics, many of them really consist of the replies that say "No, that's also correct but.."