¿Y cuánto tiempo tomará antes de que nadie ______ aprender idiomas porque todos tienen traductores en el bolsillo?
Meaning: bother
Incorrect
The correct answer is “se moleste en”
Obviously, a word is not a phrase; so the AI asks for a word and always always grades me wrong because I don’t put in a phrase, instead? Is this a joke? Solution? Always show THREE blanks if you want 3 words. And – in general – any fill-in-the-blank question should always (or, at least, by default) show EXACTLY how many characters are in the answer. So: “.. ……. ..” If the person gets this consistently wrong, maybe start filling in some of the letters for them on the re-test? Otherwise, IMO, fill in the blank is not a very efficient way for most people to learn: 99% frustration and 1% payback. As a bonus prize, since you are using Ai, more than one answer ought to be accepted if it translates the same.
The good news? I’ve tried working with the Ai on my new ipad, with voice translation enabled. So if anyone is interested in learning how to SPEAK a foreign language, instead of more interested in learning how to SPELL, they might want to switch to using a speech to text device to do their interactions. It would be nicer, yet, if the speech recognition was built into the readlang interface. Maybe a good “paid” plan perquisite?
I’ve been waiting for someone to comment on suggestion. But maybe my title put them off. ;-/ FWIW, I’m 100% ethnic Polish. Maybe that will make it okay to laugh?
Maybe in this case the entire perimeter could be considered the corner.
There was another thread where the issue of showing the number of blanks came up. It seemed like knowing the exact letter count could make sense if you’re typing the answers, but not if it’s multiple choice, as then it is probably spoiling the answer.
I’m not sure what to think of the idea that answers that mean the same thing as the expected answer should be accepted. Wouldn’t that allow you to avoid learning a term as long as you know an equivalent? Not the end of the world I guess.
Upon further reflection, I think you are right and the best, most straightforward way is to just require the entire phrase, with dashed underscores representing letters, and different words in a phrase clearly separated by spaces. The “hint” could then be changed to fill in some of the letters. Maybe allowing the user to ask for another “hint” (another letter) if they still can’t figure out what, exactly, to type. What do you think of that?
I won’t have too strong of an opinion because I always use multiple choice (Blitz no typing). But if I were typing I could see the value of partial hints. I’m not sure if answering right vs. wrong has an effect on the SRS algorithm, but if so, maybe requiring a hint to answer right should count as “less” right.