Language Learning

cross-posted from: https://lemmy.world/post/39170712

Anki is an open-source flashcard app for Windows, Linux, Mac OSX with versions also available for Android and iOS. Unfortunately, iOS version costs $25, but all other versions are free.

Anki is a self-graded flashcard program / app. This makes it a combination quiz-app + timer system. Unlike Duolingo or other programs, Anki entirely relies upon self-grading, but this is more than sufficient for study.

Anki grabs the top cards from a deck (defaulting to 20 new cards per day. Feel free to customize this to whatever fits your needs best). Then each day, it grabs "scheduled review" cards + shuffles in the new cards, and shows you them one at a time. Once a card is shown to you, you the user click a button to reveal the other side.

After the flip, Anki asks you to self-grade yourself on your performance. "Again" means you grade yourself as "incorrect", and Anki will remember this mistake. Because you were "incorrect" on this card, Anki will show you the card again very soon.

If you choose one of the three "correct" scores (labeled "Hard", "Good" and "Easy"), Anki remembers that you've answered correctly, and will schedule the card some time in the future. I'll get to the difference of the three scores later, but consider all three to just be "correct" for now.

The precise time is calculated based on how well Anki thinks you know the card. If you know the card well, "Good" might schedule the card to be reviewed 1 month from now, but if you've made a lot of mistakes with a particular card, then that card will likely be reviewed 1 or 2 days from now. Its all data collected on a per-card basis.

Above is an example screenshot of Anki's memory: every single self-graded score is remembered on every single card, as well as the date and time of each score.

As such, Anki is a system of spaced repetition. The "better" you are with some cards, the less you see them. The "worse" you are with other cards, the more Anki shows you those particular cards you keep making mistakes with. Timer + self-grading == you only see the cards you're doing bad with, while Anki hides the cards you are doing good with.

The Algorithm

FSRS is a new experimental algorithm Anki is using. There's been 6 versions (FSRS-1, -2, -3,... and of course FSRS 6 today). Fortunately, the overall gist has been the same for all 6 versions. Alas, its a lot of blogposts and technical math that's far too nerdy for most people https://github.com/open-spaced-repetition/fsrs4anki/wiki/The-Algorithm. For the math nerds who want to learn the algorithm, study away. But I'll attempt to do a simpler "translation".

Before we get started, click on your deck's preferences and scroll down to the FSRS button. Ensure it is on.

FSRS is simply three pieces of memory being applied to each and every "card" in your Anki decks. Every single card will try to figure out "R", "S" and "D". R is the probability that you've forgotten a card each day. The longer a card goes without being shown, the worse-and-worse "R" gets (this is the value Anki uses to determine when to repeat a card to you, it wants to show you a card before you've forgotten, but after enough time that you had a chance to forget, defaulting to 10% chance of forgetting).

Every single card tracked by Anki has this "forgetting" curve, primarily defined by the "R" aka Retention variable.

The theory is: if you show a card too often, you never really test your long-term memory. Furthermore, its too much extra work to review so many cards. By waiting days, weeks, or months before showing you a card again, Anki saves you time by not overly-reviewing cards you already know the information of. Furthermore, studies have shown that showing you information "right as you are forgetting about it" is the best way to remember (!!!). Any sooner, and you really aren't learning too well, but instead just temporarily holding things in your short-term or medium-term memory.

"S" stands for Stability. The more "stable" a card is, the longer Anki-FSRS thinks it can stay in your memory memory without review. Most "new" cards are assumed to be forgotten about within a day by default. However, as you get the card "correct" over-and-over again, Anki-FSRS will increase stability, thereby causing the longer review intervals. (Maybe showing you a card once every 3 days, then 7 days, then 1.5 months, then 3 months....).

"D" stands for Difficulty. The more times you get a card wrong (ie: when you click the "Again" button), the worse Difficulty gets. Anki-FSRS remembers that some cards are harder for you to remember... in particular the ones you keep getting wrong.

Even if you get a high-difficulty card correct multiple times, Anki "remembers" that you have been forgetting this card, and will show it to you again sooner. Ex: by default Anki will mature a card within 7x correct answers in a row. However, if a card is "difficult", Anki will keep showing you that card 10x, 15x or more, knowing that you need the extra practice.

Or in more math-nerd terms, "Difficulty" is the derivative of stability. The change-of-stability is determined by the "Difficulty" of a card.

Hard / Good / Easy

Hard / Good / Easy all count as correct (ie: increases the stability of Anki-FSRS), but will do different things to your Difficulty score.

"Good" is the default, and Anki recommends that users hit the "Good" button 80%+ of the time. Lets pretend that a particular "Good" answer will result in 1-month timer for a particular card...

"Easy" basically is telling Anki that you don't want to practice with this card anymore (ie: low-difficulty card). After clicking "Easy", instead of taking a 1-month timer... Anki will likely choose a 1.5-month or 2-month timer on the card.

"Hard" is telling Anki that you want extra practice with this card. It increases difficulty, despite increasing stability. You'll see this card again more-and-more in the future. Instead of 1-month timer, Anki might show you the card again within 2-weeks.

Where Anki fits in language learning

Anki was originally developed to help its original programmer learn Japanese. Its not an end-all be-all app however. Anki is only a piece of any language-learner. You must also buy grammar / theory books, as well as write regularly in the new language... speaking and listening and more.

Nonetheless, "Anki" is your cudgel. A brute-force method to try to force vocabulary words into your brain through raw force. You'll likely never gain mastery of the words through Anki... but you can at least become a beginner and learn how to start reading. There's literally thousands, if not tens-of-thousands of words you must learn to become proficient in a language. And that's spelling, grammar usage (gender / der/das/die in German, or maybe conjugation rules and pluralization rules), definitions and more!!

In all cases, Anki can be used as a way to force this information into your brain, getting it ready so that those words can "begin to be learned" when you watch TV, listen to a foreign language podcast or hear those words in a song.

Yes, Anki isn't enough. But Anki is a great tool to get you started. And getting started is sometimes the hardest step for many people.

Remember: 1000 words is beginner level (near 1st grade level understanding), while 10,000 words is roughly high school level. If you wish to be seen as a competent adult in a new language, you must figure out a system to reach those 10,000+ words known. 10,000 words sounds like a lot in isolation... especially because true mastery of 10,000 words includes spelling, grammar (pluralization/conjugation/gender), meaning, and pronunciation. But think about it: 10,000 words is merely 14 words per day for 2-years. Plenty of people have used Anki to jumpstart that kind of long-term forced-learning of words.

As someone who loves both coding and learning Japanese, I’ve always wished there was an open-source, truly free tool for learning Japanese, kind of like what Monkeytype is in the typing community (fun fact: we actually have 2 Monkeytype devs on board with us now!)

Unfortunately, most language learning apps these days are either paid or closed-source, and the few free ones that are still out there haven’t really been kept up to date. I felt like that left a gap for people who just want a straightforward, open-source, high-quality learning tool that isn’t trying to milk them and/or sell them something.

That being said, I didn’t want to just make another “me too” language app just for the sake of creating one. There needed to be something special about it. That’s when I thought: why not truly hit it home and do something no other language learning app has done by adding tons of color themes, fonts and an extremely fun and customizable experience, as a little tribute to the vibe that inspired me in the first place, Monkeytype.

So, that’s what I’m doing now. We’ve already hit half a thousand stars on GitHub and reached thousands of Japanese learners worldwide, and we’re looking to grow our forever free, open-source platform even more.

Why? Because Japanese learners and weebs deserve a free and genuinely fun learning experience too.

Live demo: https://kanadojo.com/

If you wanna make our day by dropping us a star or even contributing, then you can do so here: https://github.com/lingdojo/kana-dojo ^^

どもありがとうございます！

I'm mentally preparing to complain to a pigeon feeder in my target language. Wish me luck.

(the pigeons are filling up my window sill with shit and have learned to wait for the feeder early in the morning, waking me up way before my actual alarm)

Thanks to all the people here and for all the encourging words I went and tried again to have a chat in my target language and it went well! :)

I feel a bit discouraged. Tried speaking and it didn't go very well. Think I'll stop trying to talk in day to day activities because it really took the fun out of it.

cross-posted from: https://lemmy.world/post/38090293

This is a collection of Japanese lessons that don't rely on translation. Instead, it shows a picture to establish a concept, then builds on that.

It's based on the ideas behind Lingue Latina Per Se Illustrata (LLPSI).

I don't know how useful this is compared to other methods, but I think it's a neat thing to check out.

Do you do your flashcard reviews every day? Take weekends off? Maybe you scale your rate of new cards down instead when you need a break?

What's your flashcard pace like?

Do you know how people of your target language do things differently in casual text messages/on the internet?

A good example: russians often drop the double dot from smiling face )))) vs. :)))) or how in some middle-eastern languages they write hhhhhhh instead of the "hahaha".

Let's hope I actually use it now lol.

cross-posted from: https://piefed.social/c/LanguageLearning/p/1405694/list-of-common-misconceptions-about-language-learning-wikipedia

You thought I wouldn't ask, didn't you? Better share what you've got on your mind, or else

I'm so sorry for late thread - been a bit overwhelmed last week. But worry not! The vocabulary must grow.

I can't believe how fast this week went. Did some German vocabulary this week and practiced speaking but that did not go as well as I had hoped. Still better than nothing, though. Grammar is still my enemy.

Found recently a feed for police reports from a Norwegian county, @politietiagderlogg@activitypub.stigatle.no

I personally don't have any opinions on that county, so news on it would be by default useless to me. But as they're pretty straight to the point, short, and on a rather narrowed down subject, it's being pretty helpful for studying vocabulary, something that at least for me is the hardest part in language learning.

Introduction given, would anyone have any recommendations on Japanese, Korean and/or German feeds that follow that general idea? Not necessarily on the fediverse, though if not, at least in a way I can easily follow, such as through RSS. And while I gave a police report log as example, subject doesn't need to be the same.

Thanks in advance!

Thought I'd share them.

Some are rather niche, some you need to be a bit loose on defining as offline (Yomiwa and HanYou), and some only had physical releases.
But since some of those are older than me and I still use them, maybe others can find use in the information too.

And with how many potential languages combinations there are, plus with how degraded DRM-free software culture had become, finding those handful few below was a task. Any other suggestions?

from korean_romanizer.romanizer import Romanizer
text = input()
romanizer = Romanizer(text)
print()
print(romanizer.romanize())

Very late, apologies.