"We are resting by the Austrian mountains."
Translation:Az osztrák hegyeknél pihenünk.
9 CommentsThis discussion is locked.
What they mean is that they typed the correct, default solution, but the message for another correct solution popped up listing the exact solution they typed.
I just had that happen — I typed the default solution precisely as it's listed, yet I got a message telling me that my solution was another correct solution...I've had this happen more than once with this course.
I've seen that as well (together with miscapitalisations where the second word of the sentence was also capitalised or the first one wasn't), and I think I have an idea where those come from.
And if it is what I think it is, it's because the alternative would be quite a bit more work for the course maintainers.
(It's connected to how sentence templates containing alternatives get turned into multiple sentences by the Duolingo framework. I can elaborate if you're curious.)
I've seen all of those instances as well.
I can imagine it does have something to do with creating alternatives, though I've never really heard of the Duolingo framework using templates to create more sentences. :o
It would be appreciated if you could explain it a little. ^_^
Incidentally, I believe that multiple-choice questions ("pick all correct translations") come from those alternatives that are marked "best".
So if there is only one "best" sentence, such multiple-choice exercises will have that one sentence plus two distractors.
If there are multiple sentences marked "best", multiple-choice exercises might feature two or three correct sentences and then only one distractor or none.
If this is correct, a corollary is that "best" sentences should NOT include alternative words that are accepted but not taught in the course (e.g. hotel next to szálloda). If such words are to be accepted, they should be in a non-best translation.
Otherwise, they may appear in a multiple-choice exercise as a mandatory correct option, leading to cries of "we were never taught this word; how were we supposed to know that this sentence is also correct?"
(And similarly with advanced grammar rules.)
The way the Incubator works is that it lets you create sentences with multiple alternatives using a shortcut syntax
[abc/defgh/ijklmno]; such a sentence gets turned into several actual sentences (in this case, one with
abc, one with
defgh, and one with
ijklmno). You can also have an empty alternative by putting a slash at the beginning or end, e.g. to translate Turkish kitaplar into
[the/] books (i.e. either "the books" or just "books").
If you have multiple such alternations in a sentence template, you get even more sentences. For example, A pizsama nagy, a nadrág pedig kicsi. might turn into The [pajamas/pyjamas] are [large/big], [and/] the [trousers/pants] are [small/little]., which would give you 32(!) sentences, including The pajamas are large, and the trousers are small., The pyjamas are big, the pants are little., The pajamas are big, the trousers are little. etc. etc. There are five alternations in this example, each of which has two possibilities, so you have 2×2×2×2×2 = 32 actual sentences that come out of that hypothetical template.
For every sentence, there are two kinds of translations. I'm not sure what the official names are; I call the one kind "best" translations. They're marked with a blue star in the Incubator interface. There can only be a maximum of 10 (I think) sentences which are marked as being a "best" translation.
This number can come about either because you have 10 individual sentences each with a blue star, or a single template that works out to 10 sentences because of alternations (say, one alternation with two choices and another with five choices = 2×5 = 10 sentences), or a combination (e.g. one template with an alternation of four choices + one template with an alternation of five choices = 9 sentences).
The first such marked sentence is, I believe, the one you see in sentence discussions. If the first template contains alternations, then the one shown is the one produced by always picking the first option in each alternation.
Since you can only have a limited number of "best" translations, it's often best to keep alternations down in them; perhaps even to omit them entirely and only have literal sentences at the top. For example, for our example sentence above, we might have The pajamas are large, and the pants are small. as the "best" translation.
Now the problem is that this produces two identical sentences! The pajamas are large, and the pants are small. is accepted because it's given explicitly as a sentence (the best translation), and it's also accepted because it's one of the 32 sentences that comes out of the template.
The system won't let you enter two identical templates, but the dupe checker doesn't check whether two templates could produce an identical sentence once you've expanded all the alternations.
So what happens is that if someone enters The pajamas are large, and the pants are small., the system may tell the user that that is a correct solution (one generated from the template), and that there is another correct solution called The pajamese are large, and the pants are small. (the "best" solution). Leaving the user to scratch their head, wondering what the difference is. (I believe that the "another correct solution" shown is always one of the ones marked "best", probably even the first such one, and only appears if [the system thinks] you entered an alternative that is not the best one.)
So how do you avoid this?
Well, imagine that the alternatives are My [father/dad/daddy] said [hello/hi]. and your best translation is My father said hi.
Then you could add these sentences:
- My father said hi. [marked as "best".]
- My [father/dad/daddy] said hello.
- My [dad/daddy] said hi.
If you think those through, that should give you a total of six sentences (1+3+2), the same number (and the same sentences!) as the version My [father/dad/daddy] said [hello/hi]. -- but each of them only once, either as the "best" translation or as a non-best translation.
But this is a fairly simple example, yet you still have to think fairly carefully about how to construct it so that it matches exactly the versions you want, without matching anything you don't want. And with the pajamas/pants example, you're looking at something like:
- The pajamas are large, and the pants are small. ["best"]
- The [pajamas/pyjamas] are [large/big], the [pants/trousers] are [small/little].
- The pyjamas are [large/big], [and/] the [pants/trousers] are little.
- The pyjamas are [large/big], [and/] the trousers are small.
- The pyjamas are big, and the [pants/trousers] are [small/little].
- The pajamas are big, and the pants are small.
That's 1+16+8+4+2+1 = 32... which maybe are the same as the original 32 versions. Are you sure, though? If not, what do you have to tweak to get it right?
Now imagine that after you've finally got it just right (there's a button that shows you all the generated sentences that you can use to check whether all the right ones are there and no bad ones show up, and you can enter a sample sentence and get told whether that matches any of the templates or not), people complain that in addition to "pants" and "trousers", you should also include "slacks". So you think about how you can modify those six separate templates such that you also get "slacks" in the right places to come up with 2×2×2×2×3 = 48 new sentences. Or maybe you need seven templates now to get it right?
This may be the point where you say, "Hang the users; I'm going to use just The pajamas are large, and the pants are small. [best] + The [pajamas/pyjamas] are [large/big], [and/] the [pants/trousers/slacks] as [small/little].. Two versions of the sentence: one best and one template with all the non-best versions."
Now, another thing is that the template system is a bit smart with alternations that come at the beginning of a sentence.
If the first element of the alternation starts with a capital letter and contains an empty element (a slash at the end, or two consecutive slashes in the middle), then the sentence produced with the empty version will capitalise the next word instead.
This means that you can translate Az autók régiek as [The/] cars are old and get two sentences, The cars are old and Cars are old, with the second one capitalising "cars" automatically.
But if you don't know about this rule, you might do something like [The/] Cars are old, which gives you The Cars are old and Cars are old.
Or if you typed it as [/The] cars are old, you might get (I'm not so sure about this one) The cars are old plus cars are old (with an uncapitalised car as the first alternative was not capitalised).
I think this sort of thing may lead to odd capitalisations at the beginning of sentences.
Or perhaps someone had Egy könyvet olvásunk. and then someone decided to add an optional [/Mi] to the beginning without lowercasing the Egy which was now no longer necessarily at the beginning of the sentence.
...that was kind of long, but I hope it shed a bit of insight into how things work "at the back".