back
363 comments
As someone who knows four languages[1] (picked every single one up during childhood) and is currently learning Sanskrit, I have to say that Krashen's input hypothesis and Orberg's Lingva Latina is probably the way to go if you are learning languages as an adult.

The direct teaching method works but is time-consuming and generally used for languages that lead to an occupation, viz. English. The grammar translation method is a waste of time. It might satisfy your intellectual curiosity about the structure of the language but you won't be able to make yourself understood after a lifetime of study. I wonder at the sheer lunacy of dumping thousands of random sentences into your lap and translating it from one language to another.

After a year and a half of false starts, I started reading a couple of Sanskrit stories every day. Because the context is maintained across the story, your brain starts recognizing patterns in sentences. You keep reading sentences like

sarvē janāḥ kāryaṁ kurvanti

sarvē janāḥ gacchanti

sarvē janāḥ namanti

and you automatically associate sarvē (all) with janāḥ (people) without needing to know the declension of those words. This applies to the cases as well.

To be able to converse about or understand a wide variety of topics, you will eventually have to move beyond stories due to restrictions on the tense/aspect/moods you encounter as a result of the nature of the material. But that is doable.

[1] Much of India is bilingual. A substantial minority might know four or more languages due to the many mother and father tongues and heavy internal migration across the states (whose boundaries were drawn on linguistic lines post-independence)

I built a tool[0] that gives you constant input at your level as you browse the web, so you don't need to take time out of your day. You can just learn a little as you browse, and let it compound over time.

It works by estimating the difficulty of English sentences, then translating ones at your level into your target language.

[0] https://nuenki.app

Duolinguo tries to follow the input hypothesis. For instance, it barely teaches any grammar but simply asks its users to translate sentences. Unfortunately that's very ineffective. Compared to reading or watching live conversations, the amount of input in unit time on Duolingo is too little. In the meantime the sentences lack sufficient context for Duolingo users to build up intuitive understanding of phrases. Take Duolingo Japanese for English speakers, for instance, it's really hard to learn the meaning and usage of Hiragana words in those short sentences.

That said, I still do about 10 minutes of Duolingo every day, just as a kick start of my daily language-learning routine. It's also an effortless way for me to pick up a few new words on a daily basis. Somehow once I did that, I have more drive to do more comprehensive input by watching Youtube videos or reading some readers.

> I wonder at the sheer lunacy of dumping thousands of random sentences into your lap and translating it from one language to another.

I don't get it why everyone seems to think that translation exercises in a foreign language learning course such as Duolingo absolutely MUST result in a comprehension-less memorization process, which must be doomed to fail sooner or later, since memorization alone might not really contribute to the capability to build new combinations of memorized words.

From my experience with Duolingo, it all depends on how a learner approaches translation exercises. If you just keep sprinting through such exercises, in a sense, mindlessly, without asking yourself how each new sentence really differs from the ones you have already seen, then yes, IMHO you are likely to fail.

However, if you keep investigating, on your own accord (for example, by using an LLM) the underlying REASONS as to why each new sentence really differs from the ones you have already seen (i.e., grammar), then no, IMHO you will indeed learn how to build new language constructs and thus use the actual language.

I think the trick is to push yourself and attempt - as soon as you can - to ignore sentence "building blocks", "missing words" and "hints" provided by Duolingo and always try to build an answer to every exercise entirely from scratch in your head. That forces your brain to understand what is really going on and create a "set of rules" for using a language as opposed to only memorizing a "set of samples" of a language.

I also don't mind the "gamification" of the learning process: it allows a learner to expect more out of himself or herself by watching it not to carelessly lose the "hearts" exercise currency, by trying to earn the "gems" bonus exercise currency, by comparing himself or herself against his or her peers through leagues and leaderboards and, the last but not least, by continuing to learn every single day because of his or her running "learning streak".

Duolingo can give you only as much as you decide to get out of yourself - as is the case with any other kind of foreign language learning course. Effortless, magical learning processes simply do not exist.

My own experience mirrors yours. My first thought in seeing this was “…why?” Duolingo is a gamified app that feels like learning a language but actually teaches you next to nothing while driving engagement. I get why they got stuck on that path, but why copy it?
I learned Sanskrit by translating the Bhagavad Gita (https://gita.pub), and I experienced a similar jump in comprehension to what you described. At first I needed to look up every word, even the ones I'd seen many times, but eventually (after many many repetitions) I finally started having an intuitive idea of what the words and sounds meant.

It certainly makes you appreciate the unbroken oral tradition by which these enormous works of literature were passed down.

i don't know why people are taking Duolinguo and relatives as the definitive course to learn a language... they even cite at their FAQ about the need of going outside the app if you want 'fluency'

some people are quite fine learning a limited number of phrases to lurk in a country. a great part of communication among humans also happens with the body/eyes. no one needs to discuss their phD dissertation in 4 different languages

[0] https://blog.duolingo.com/can-duolingo-make-me-fluent/

edit: Duolinguo also is nice (and make a funny non-invasive joke) if you are using something like uBlock!

Ah, is Lingva Latina the one with Caecilius and his family? I had a Latin class in 7th grade and remember having a book of that same name, and I somehow remember the main father character's name. They had a dog too, I want to say his name was Cerberus, haha. "Cave canem"—"beware of dog"

Every day, we'd start class by the teacher saying "Salvete, discipuli!" to which we'd reply "Salve, magistra!"

The fact that all these years later I still remember some things from it shows its effectiveness I suppose.

In any case, in years since, I've used Pimsleur (for other languages), which is a similar "get actual language input rather than learning a set of language rules up front" method, and I like to think it's worked decently for me at least!

Orbeg's Lingva Latina is so good, especially if you use the supplemental exercises and stories as well. It's a shame it didn't catch on as much, the material for modern languages is now outdated and it seems no one is working on newer editions. Deutsch Nach der Naturmethode, Français par la Méthode Nature and English by the Nature Method are excelent at teaching the basics, but I hear criticism over the vocabulary often.
I’m very interested in Sanskrit, and working on an application to learn it (and many other languages).

If you have any interest in app based review (not courses - I specifically try to work with input) I would love to get feedback on the Sanskrit experience.

I posted a bunch of comments about it in the past few days, I don’t want to take over another apps thread, but there are so many cool languages being learned here

Very nice initiative, the language space is overcrowded with commercial offers that have an incentive to keep you locked in. Apart from LanguageTransfer there seem to be few other good offers.

That said, looking at the current offer it seems to lack the one thing Duolingo offers: Duolingo (for all its many faults and pedagogical uselessness) takes the burden of decision making away - I don't need to really think what to do next. Here I don't have this guidance - do I start with basics? Or introduction? Or something else?

Crucial in my view would be to provide a path or at least a tree to guide the user where to go. This will make it easy to jump in and get carried along.

Duolingo user here with a 4 year streak.

Duolingo is not a language teaching platform at its core. It’s a gaming platform with language as its gaming skill.

Duolingo at some point became so focused on gamification that it just became a game (I believe they hired their lead PM from Zynga).

If you’re on free version, just look at the ads you’re getting. Vast majority of the ads are for other games.

I think you can learn a language if you use Duolingo’s streak gamification as a daily motivator but use supplemental materials to actually learn.

For the curious, here's an article from the developer on why they built LibreLingo https://dev.to/kantord/why-i-built-librelingo-280o
I used Duolingo for about a year to learn Portuguese but I recently switched to just taking a course I bought on Udemy.

First let me say that Duolingo is great for learning vocabulary but unfortunately that's it's only strength. The problem I realized after starting the Udemy course is that Duolingo teaches you the words but they seldom teach sentence structure or the "glue" between all those words you learn. So you get to a place where you know a ton of words but can't hold a conversation because you don't know how to form sentences.

With that said I would still recommend Duolingo strictly for their vocabulary. I would suggest a course to supplement learning though, not to mention it's much cheaper, the entire course cost me less than a month of Duolingo Super.

The problem with duolingo is that translating a language is not the best way to learn a language. The best way is to make a connection between the concept and the word. Like rosetta stone does. An open source rosetta stone would be better, at least for learning vocabulary
There wasn't much to read there, but why aspire to be an alternative to Duolingo of all things? Duolingo focuses on learning by translation, basically. It's even in the name: "Duolingo". It's an utterly broken approach to learning languages, except for the very initial phase where you're getting just enough to move on to modern methods (i.e. avoid translation like the plague, to start with). Which is exactly why a comment I read somewhere said "Duolingo is for the perpetual beginner".
Cool... but nothing will ever beat Anki + Immersion. Here's one guide most Japanese learners follow: https://learnjapanese.moe/
been using Duolingo in the 10s and last year, I gave up because the course seems very repetitive. Even if I got the answer right 10 out of 10 times, the same question kept coming. It almost looks like the app is trying very hard to make me stay as long as possible, instead of study as effecient as possible.

So for a good alternative app, is there a dynamic course pace I can adapt to?

Sadly the authors of LibreLingo were last seen being lead into the back of a white van by an enormous green owl
I'm surprised more people aren't talking about the recent announcement that Duolingo is replacing contractors with AI.

https://www.theverge.com/news/657594/duolingo-ai-first-repla...

Going to plug Language Transfer again, an excellent free app that is a much better way to learn a language than the DuoLingo approach.
I like it! Really fun and fluent, though maybe the keyboard navigation (e.g. radio boxes, etc) could be improved.

I like the turtle, but maybe you want to rethink the jetpack flames from it's behind approach. Also, maybe a slight more "shiny" version, a la Duo, would match nicely.

But overall, great work !

The app could use some spinners, when actions lead to a delay. I clicked on the landing page on the only available purple action button and nothing seemed to happen. I already checked my uBlock Origin, whether it is blocking something, but it does not. Already wanted to reload the page, when finally something visually changed, and the course was loaded. Simply a little spinner/animation would make this way less confusing.

I like, that for keyboard input the special letters are given as buttons, so that I don't need to hunt for those on any US/English keyboard layout.

One thing missing is a way to report mistakes in the learning material. For example I found "Buenos dias" to be translated to "Good morning".

I'm a little surprised that Duolingo is the model someone wants to emulate because, at least for me, it just doesn't work.

Now I'm someone who has always been good at taking tests. It's a skill you can develop. At one point I got 85% in a French test knowing absolutely zero French. There are tricks such as:

- Use of punctuation can give the answer away (eg a trailing "!")

- Other questions can unintentionally give you the answer to a different question (eg it might conjugate a verb you're being asked about elsewhere);

- Questions end up being correlated. So a given question might have 2 plausible answers and that answer will also answer another question. So you can answer if one way in one and another way in the other and you're pretty likely to get one of them right;

- Multiple choice tests tend to evenly distribute answers so if you have 29 Cs in a 4-answer 100 question test already, it's less likely that a further C guess is right. Yes, people can intentionally re-weight the answers to avoid this but almost nobody does.

- For other topics like math you often get marks for each step. Depending on how that marking key works, you can often get marks writing essentially nonsense that leads to a completely wrong answer;

- When in doubt, guess something. This goes for multiple choice and written answers. Don't spend any time on it. Tests that deduct points for wrong answers are rare and you know about it beforehand.

- Apply probability. So in a 100 question 4 answer multiple choice test where you have a 50% chance of knowing the answer, you should really get 75-80% on that test just from eliminating obviously wrong answers and simply guessing the rest.

My point is that you can't really turn this off once you learn it so I can pretty much guess my way through any Duolingo questions and that means I don't learn anything.

Even when you have to assemble words into a sentence, the answer is pretty obvious and it can get even more obvious in other languages (eg nouns in German are capitalized).

I think I did Spanish Duolingo almost every day for a year and remember none of it.

It sucks how Duolingo has gotten so much worse over the years.

It used to be great when it had the grammar notes and discussion forums and comments, and you could actually finish the course and have some recognition.

Now it's just all too game-like and all based around maintaining streaks rather than learning.

Unfortunately some other apps have started to copy this model too like HelloChinese.

It's a bad idea to imitate Duolingo, which has become VC without a purpose.

The gimmick behind Duolingo was that there were so many things online and in the world that needed to be translated, so training people to learn languages while translating them was a win-win. We don't really need humans to translate written material anymore (esp with AI advances), and they never seemed to find a business model for that anyway.

Since the gimmick is gone, it's just a generic language learning app with unimpressive results. And that still uses primitive spaced repetition algorithms. The bottom fell out. But since Duolingo had attracted a ton of cash on their founders rep from reCaptcha, it zombies on.

I've had my account since the beta, and while I think it's good because it exposes people to a ton of words and utterances in their target language which they will hopefully roll around in their mouths, that's like step 1 in learning a language. Anecdotally, I had to abandon Duolingo entirely in order to learn Spanish; and not for a class or tutoring, but for their competitors both online and traditional.

Techniques in language learning seem to be advancing quickly (like with spaced repetition, TPRS, and Krashen-inspired stuff), but Duolingo seems to be studiously ignoring them all, and plowing on doing the same thing. I think they should ditch everything but the cartoons, which are cute. But their base gets outraged whenever they change anything because Duolingo's changes were made in order to shift to getting revenue from the users rather than from "translation," so the users do not trust them.

So Duolingo really have nothing but cute cartoons and a brand name. LibreLingo looks like they have cartoons, too. Other than those, there's nothing to distinguish Librelingo from any other Spanish-learning website.

The problem is that Duolingo optimises for time spent on the app, not for progress in the language. The majority of experienced language learners do not recommend it.
If I’m going to spend a thousand hours learning a new language, I’m willing to pay for professional study material.
I'm sure this is a wonderful project with talented people behind it, and what I'm going to say isn't a criticism of this project in particular.

But. I'm always a little disappointed when I see a project that's Libre[something proprietary]. It's always a wonky copy, where the selling point is that it's a free version of something, rather than a better version of something. The only people who are going to use it are those who care more about the fact that it's free and Libre than they do about a good learning experience [0]. Everyone else will just use Duolingo. And that's fine if the goal is for it to be a programming exercise, but it's a limiting one.

Instead of making a knockoff of Duolingo, which clearly been eaten by the pressure to drive engagement and MAU, why not use time and energy to explore different or more radical ways of online pedagogy free from commercial pressures? It's harder than copying something, but the results could be much more worthwhile. [1]

---

[0] This is why Mastodon will never go mainstream, because it's built by and for people who care more about decentralisation than they do about creating a first-class microblogging experience. The friction points that deter the mainstream are acceptable for the true believers because for them the benefits are worth it.

[1] This is also my problem with Linux desktop environments. The desktop war was won by Microsoft 30 years ago and the desktop died as the primary computing paradigm in 2007. Yet Linux desktops are still fighting the last battle - so much time and effort is poured into them, yet they still don't work right (Wayland is how old now?) and are basically just wonkier versions of macOS or Windows.

Surely that time and effort could be spent on investigating new ways to interact with computers - why is the desktop metaphor still the best we've got, nearly 60 years after it was first invented?

If I could change one thing about Duolingo, it would be to allow the user to turn off all the gamification completely. I don't care about fake internet gems, knowing how to speak Chinese (or whatever) is it's own reward, so stop wasting my time with bs!

Nevertheless, Duolingo is an amazing and convenient starting point for unlocking the learning of new languages.

Make your way through the entire course as fast as you can, while also listening to music, talking to people, talking to chatGPT, reading books, etc in the target language as soon as you can manage.

Protip: learn your 3rd language using your second as the language of instruction.

Can anyone recommend a solid resource for learning Tagalog?

Being a far less popular language than the standard "big boys" that most apps, web sites, books etc tend to offer, it's been a lot of false starts for me or simply feeling a bit lost when a resource throws me directly into a scene to learn dialog without having any of the foundational knowledge first.

I plan to spend a year or two (at least) in the Philippines in the not too distant future. While most Filipinos understand English, I feel like learning at least some Tagalog would go a long way in fitting in and feeling less like an outsider.

Great to see a FOSS app for language learning! Kudos.

However, I think apps that focus on one particular language and how to learn that language are better than a one-size-fits-all approach like Duolingo. The structure and grammar of languages like Russian, Japanese, Chinese, and French (I've learned all 4) are all significantly different from each other. Or at least different approaches for language groups (French and Spanish, which I also speak, are similar enough to warrant the same approach).

This space seems like one of those areas where it would be really hard to break in because their whole selling point is having had hundreds or thousands of people record and annotate an enormous amount of voice input, which I assume has to be hand polished for every single exercise?

I'm sure some part of it could be automated these days, or some parts even use voice synthesis, but I'm sure it would take basically an army of people hand-crafting it for the experience not to be very janky in the end.

I can recommend https://polski.info for Polish. Not FOSS, but at least non-commercial.
Nice to see this pop up, not that I mind giving Duo money every month for my kids account.

Still looking for DuoLingo for actual programming... python etc... Specifically for elementary school kids... I know it's out there... Im getting closer...

I know this is a false statement but it would be so easy for DuoLingo to add Python along side their Math and Music betas!!

Please Duo hear my prayers...

Anyone as experience with feedback on https://www.rocketlanguages.com/ and https://babbel.com/ ?

I’m mostly interested in speaking out loud skills, and those two have voice recognition it seems.

Has anyone tried both LibreLingo and Duolingo? I'm curious if the open-source approach makes learning feel more natural?
The signup button just spins indefinitely :(
Yeah, I've used Duolingo and always end up doing the same stuff, but it's way too easy to just tap through and not actually learn anything real. Got me thinking - you think anything can really replace just talking to people and living with the language day to day?
Not FOSS, but for anyone who may find it useful, I launched a language learning project I've been hacking away on for a few months. Any feedback most welcome! https://langsesh.com
Related ongoing thread:

Duolingo will replace contract workers with AI - https://news.ycombinator.com/item?id=43827978 - April 2025 (43 comments)

Anyone have any suggestions for learning Korean? I have Hangul characters down and know some words/phrases from my partner, but would like to dive in a bit more.
Hope this takes off! Early Duolingo was very community focused and lots of fun, and proves people are really motivated to participate if you make the UI easy enough.
But why? These apps are ineffective. If you want to learn a language, don't waste your time on Duolingo or this...
All language apps are destined to become essentially an SRS app, which at that point you might as well just use an anki.
Really hope they can do something about the UX; well built OSS generally lacks good UI/UX.