[ngokevin]

I'm trying a language learning startup and surveyed a lot of people. The response is universal that everyone's tried Duolingo, and it didn't work for them in actually learning, despite monster streaks.

I think it's masterful as a dopamine machine and employing habit building tactics. Perhaps it's a good icebreaker for people in language learning, as long as people are resourceful enough to look beyond Duolingo.

[thesausageking]

Duolingo has worked great for me. I had learned Spanish a decade ago and completely forgotten it. In a few months of consistent practice, I learned enough to get around. And after 6-9 months, I could have basic conversations.

I was never good at learning languages and Duolingo has been the best system for me. I like the paced repetition it does. Also, that it has you practice reading, writing, saying, and listening.

[dorchadas]

> I had learned Spanish a decade ago and completely forgotten it.

I would argue this is why it worked well for you. You already knew the language and just needed to revise, which is something I would argue DL does do good at, especially the web version.

> Also, that it has you practice reading, writing, saying, and listening.

The thing is, in most courses (the home-developed ones with stories might be exceptions), you don't practice those skills. You learn to translate stuff from the target language to English; that's not reading. You don't learn to read texts and interpret them in the language. The writing is awful as well, since it's just reverse translation, not actually responding to a prompt or natural conversation.

The speaking leaves a lot to be desired, unless it's massively improved since I last tried it. I once said "blah" on the Spanish course and it was accepted as correct; I've also had it accept background noise before too. And the same with listening -- it's a TTS, you don't actually practice listening to native speakers and understanding what's being said with comprehension questions, etc.

It doesn't really teach the four skills as they would be applied in the real world at all. Not to mention that default on the app has you click words in the proper order instead of actually recalling them. That's another huge negative.

[linsomniac]

>You don't learn to read texts and interpret them in the language.

Are you sure? I regularly come across questions that involve a few sentences, a question, and a multi-choice answer. All in language, and some of them are somewhat tricky. Something along the lines of: "Juan's is with his girlfriend. Her name is Antonia. She is buying a green dress. Where is Antonia?" "A: A restaurant. B: A clothes store C: A party."

There are also questions that are a few sentences and you have to fill in the blank, which often require understanding the context, in language: If Juan is speaking to waiter, then the place he is at is a restaurant...

>Not to mention that default on the app has you click words in the proper order instead of actually recalling them.

Not at all true. I've been using DuoLingo for less than a month and I regularly see questions that are an English sentence and a free form text box for you to type the translation into. And one of the questions for the second "Checkpoint" is: Here is a sentence in language, and a free form text box to type the response.

As far as the clicking words, I'll say I really like that as one of the components of learning, because it lets me focus on the structure of sentences without getting frustrated by typos or misremembering words or even in some cases forgetting words. The prompts of the bubbles are good and still make me remember the differences between the conjugations and the like.

I spent a semester in a class in Jr HS, and was much more frustrated and less educated than I feel after 3 weeks of solid DuoLingo. For me, it's a fun tool for establishing some confidence and understanding.

[cerved]

it's not multiple choice if there's only Juan answer

[vncecartersknee]

That's cool I'd love to hear more about it, I've actually been thinking about ways that a language app could work better. Seems like all the current research is that "immersion" in the target language is critical, an app that just played me conversations, had me read text, etc and 'creatively' formulate responses, not just simply fill in the missing word, obviously in increasing difficulty would be ideal.

[kouteiheika]

> Seems like all the current research is that "immersion" in the target language is critical

This, a hundred times this. Immersion is not only effective in getting you used to the language, but also really helps you solidify what you've learned, and is extremely motivating.

Virtually all of the people that I know in the Japanese learning community who learned the language really fast heavily immersed in a lot of content.

In my own language learning app that I'm making I'm explicitly targeting a feedback loop which looks something like this:

1) You pick an easy show/book/etc. to consume. (I have a list ranked according to difficulty, generated through machine learning.)

2) You prelearn the vocabulary used within using an SRS.

3) You consume it.

4) You continue maintaining your vocabulary through SRS, so the app knows which words you're supposed to know.

5) The app recommends you the next show/book/etc. to consume based on what you've already learned.

6) Rinse and repeat. Gradually the whole thing snowballs and you have to learn less and less for each new work you consume.

I personally believe this to be the most effective way of learning a language, and I'm already seeing some of my users having a lot of success with it.

[davidzweig]

We are moving 'Language Learning with Netflix' in this direction. There's some stuff to see already: https://www.languagereactor.com/ (my email is in my profile, I'd be happy to hear from you :)

[dorchadas]

If you could do step one that wasn't based entirely on anime I'd love it. That's always been my biggest hurdle with the Japanese learning community -- almost all recommendations seem to be anime, and I'm just not that interested.

[kouteiheika]

I also have Japanese dramas, light novels, a few normal novels, visual novels and I recently started adding YouTube videos, so it's definitely not only anime.

[dorchadas]

Perfect, thank you! Please let me know when you release, I'd definitely be interested in it.

[kouteiheika]

Well, it's already publicly available. (:

Here's the link in case you want to take a look: https://jpdb.io

Just a fair bit of warning: it still has ways go to, so don't expect a perfectly polished experienced. That said, in spite of its rough edges I do already have users who use it do hundreds of cards per week and successfully immerse, so it is usable.

[k__]

I had a somewhat similar idea once.

You share Twitter accounts with a service, and it would use the tweets of that account to generate a language course.

I was hoping that this would lead to learning something that I could use right away.

[monkeybutton]

Immersion provides the strongest motivation to learn and to try. When you want something or have a problem that needs to be solved and the other party doesn't speak English, you're going to be very motivated. The 'dopamine hit' from getting it right and communicating successfully in another language is also way better than from an app/game. The results are just so much more real and tangible.

[Jare]

Immersion also creates connections between the language and other stimuli (visual, emotional, etc). Connections are powerful for creating reliable memories.

[jcranberry]

My "theory" is that learning a language is like learning any other discipline to a high degree of fluency. If you study a couple hours a week even if you don't forget what you learned in short order it's something which requires a period of immersion for you to be able to recall it as well as the day you learned it. To be able to expound at length on a topic at the drop of a hat you essentially need to dedicate a few hours a day every day. I think languages work the same way.

[ngokevin]

I'm targeting couples that are bilingual that want to learn each other's languages. Immersion is difficult to achieve alone. So I'm trying to channel a native speaking resource that is around you all the time.

Right now the system is a mixture of SRS on your own, sourcing vocabulary from songs and podcasts, and combined with immersive multiplayer mini-games with the native speaking partner.

[trwired]

> The response is universal that everyone's tried Duolingo, and it didn't work for them in actually learning, despite monster streaks.

I think this is an issue of mismatched expectations. For actually getting to know a language a tool like Duolingo is too limited, as knowing a language is an umbrella term comprising multiple skills and Duo helps with only vocabulary, a bit of grammar and pronunciation to some extent (but I have no experience with that aspect). If someone starts a course thinking they'll be able to speak the language once it's over, no wonder they eventually end up disappointed. What Duolingo is excellent at, however, is kickstarting the process of learning a language you're interested in. Hard to find a better tool for that.