"The child is now sleeping" is graded as incorrect. Isn't this the same?
Another new word (hiện tại) introduced with no word list now, and no vocabulary training, or whatnot. And somehow, someway, we're supposed to know this when we get tested? And it's only given in an audio test to boot! (If it was given as a reading test, we could at least google it to look it up).
Once again--trial and error learning is no way to learn a language any more than flying an airplane without being given any preparation or instruction and crashing it is a good way to learn how to fly. I find DL's methodology very frustrating and demotivating.
It's a big debate, Stewart.. But science has now demonstrated that it is more efficient to learn through trial and error than through rote learning of vocabulary lists. Every error is a learning opportunity. I understand you find this frustrating, but be gentle with your self. Every time you make a mistake, say 'thanks for the learning' and keep smiling ;-)
As a scientist myself ;) I differ on that. Like, do we teach physics or chemistry or electronics by sending students into a lab and letting them find out by breaking things (the best case) or blowing things including themselves up (worst case) ? Do we teach kids how to drive by handing them car keys and telling them 'have at it'? No classroom I've been in is it considered fair to test students on things not covered previously.
Learning all these things and more involve you learning the rules before you begin practice, and in that practice you begin with examples where the rules are followed, not the cases where they are not. Languages too have rules, and we should start by learning the rules. I accept that in all languages there are times you have to learn something that breaks the rules "just because that's what usage is" but these should be taught too as exceptions where the rules get broken.
While it is true that error is a learning opportunity, that's only if there is some rule in place that is explained, or if in this case, the usual rule does not apply. Without that explanation and framework, there is no learning in trial-and-error. That is entirely lacking here.