memcode
memcode copied to clipboard
lakesare
Automate the creation of cloze-deletion flashcards
When we create cloze deletion flashcards for some word we don't know, e.g. "grit", we frequently go through the same algorithm manually:
- Go to lexico.com: https://www.lexico.com/definition/grit
- Copy a good example sentence
- Copy the definition
- Finally create the flashcard.
Is it possible to automate these tasks via the dictionary api?
I tried an app called Clozemaster and it works like you suggested, it's pretty good, but freemium. For Korean it sentence mines Tatoeba.
The best feature is actually cloze cards with TTS, where you have to listen to the card's audio and input it. Really works your listening comprehension in context.
So after using it, I think this is a great idea. Even better if you can just import a bunch of cards without even choosing the word, like by certain parameters like difficulty. Then you can just make a deck, import a thousand sentences at your level and get cracking right away
Yeah text-to-speech is very good for new languages; and already-existing flashcards are also pretty important for new languages, memcode lacks in that (I had to learn some german over the summer and it was evident memcode wasn't up to speed here; only good when you're already rather advanced).
The first suggest (getting to crunch words right away) is could be implemented simply by creating courses. For example, we could have an admin-approved 500-most-used-words courses per each language.
The second suggestion (TTS) is something I could implement, but would require some time/funding.
The best part about Tatoeba is not the words, it's the sentences. You need to be a native to decide which sentences are culturally significant.
Example: Good morning is 좋은 아침이에요, but it is not used as a greeting in Korean. In the morning you still wish someone tranquility, just like any other time.
The sentence actually means the morning is good, so from the grammar standpoint it's not wrong.
Tatoeba also has recordings:
https://tatoeba.org/en/sentences/show/2704960
None for Korean, but that's just an issue of nobody making them.
But let's say you want to make and maintain courses for languages, what's the UI for making those first 500 cards? It's 500 for each language translated to another language, so 500 per mapping
Not words of course (imagine that!), I meant sentences with words cloze deleted/question-answered.
I think there should exist some databases with such sentences, please @ me if you know of them.
Having such courses sounds like a very good idea now that I'm thinking it over.
https://glitch.com/edit/#!/browser-voices?path=index.html%3A1%3A0
Before I forget, I got this to work on my phone. Requires enabling text to speech, but the Google one is pretty good
Nice! I worried non-english accents might be bad with a free browser api, did you check it for Korean?
Nice! I worried non-english accents might be bad with a free browser api, did you check it for Korean?
Korean sounds great, so does Chinese
There's just a weird bug where if you have language detection off it will read it in English, even if written in the Korean hangeul
this is a really cool feature for this type of card as well, since it looked up by selection (anything I select is automatically looked up), not by word - I just selected the answer and got the Chinese characters straight from wiktionary so I can relate the 차 to meaning of the character I already know, 次 which means "x times" while the whole word means more like "step by step"
the clicking part is not useful because wiktionary doesn't have the combinations with post-positions and full search just finds random articles with example sentences
although this is maybe a complete separate feature
I have an extension in my browser for this (https://addons.mozilla.org/en-US/firefox/addon/dictionary-anyvhere/?utm_source=addons.mozilla.org&utm_medium=referral&utm_content=search):
Maybe this functionality is better as an extension.