If the reason described is true (AI will replace human translators), then Duolingo’s often nonsensical example sentences in certain language courses will probably become even worse. Or perhaps it’s the other way around… but right now some language courses like Finnish are full of what feels like bad-AI generated texts. “Icelandic pony is a Viking” sort of stuff (I did not make that up)
What is the "big layoff"? 10 people? A hundred? A thousand? What percentage of total employees is it? Is there any reason to believe that they were "replaced by AI" vs just being a standard layoff like for the rest of the tech industry?
We need a better source for this than a random Reddit self post before bringing out the pitchforks.
Duolingo really fell off the map for me for two reasons:
1. They moved away from community built language courses to ones built by “experts.” To be honest, I think the “expert” ones are much worse. The phrases seem more robotic and the structure… what structure is this?
2. We can discuss diversity another day - but what’s not welcome? Literally every screen having to be some new random combination of race, gender, disability, and cultural background. It’s so forced to the point it feels, for lack of a better word, like an indoctrination. I’m not against diversity - but badly handled stuff like this is how you build resentment. If we’re going to have diversity, why not primarily show people who actually speak the language?
I'm actually really interested in why people feel this way when in these situations. It's an app with cartoon-y characters and somehow when they don't match identity expectations, people seem to feel something about that? It seems so low stakes to me, I simply don't care about them, but some people seem to be really invested in these cartoons and that means I have something to learn about how brains work.
My main issue is that it’s scattershot diversity, not a thoughtful diversity.
Scattershot diversity is what Hollywood films also are finding out causes anger. It’s when you take all the races, genders, etc., and just make sure there’s at least one of each in every movie. This also ironically cheapens diversity, because everyone doubts whether that character would have been there if it wasn’t for the race card.
Compare this with targeted, meaningful diversity. A film studio might make 4 films: One set in 1500s England, one in modern Congo, one in Brazil, one in Mongolia. The British one is almost entirely white people, the Congo one almost entirely black people, etc.
My point is: Vary up the locale, and keep the diversity accurate to the locale. If the locale is diversified enough, the diverse representation will follow with almost no complaints. It’s not like Europe has a monopoly on stories - I’m sure that every culture has stories untold that could be adapted into interesting material.
I have yet to meet anyone legitimately angry for the reasons you claim. I have met several who profess such a belief, but in every single case it is the presence of anyone not like them in the representation that they disagree with, not any actual thoughtfulness. These people always find a new goalpost to move to: "How dare they have a black person on this show?" "How dare they have a black person kiss a white person on this show?" "How dare they have a [X] on this show?" "How dare they have characters not speaking English on this show?", etc., etc.
Not being pandered to 100% of the time does not mean that someone is being repressed. These are *auxiliary graphics in a phone app*. Not even UI elements. Is this really something you consider worth getting worked up about and resenting?
It sounds like you welcome diversity, but have high standards on what folks might call "token diversity". Did I get that right? I'm not exactly the right way to phrase this, but you might say that token diversity isn't grounded with in-universe explanation, or something like that?
If that's true, how do you approach topics like sex and gender? How does one justify a character being queer, trans, &c? On one had I can see, "their identity should add to the story," but on the other hand in real life plenty of people simply are queer or trans or whatever and completely and have stories to tell that don't relate to their identity.
And what you have characterized as scattershot diversity is basically same as racial caricaturization. Is adding a heebie lussiyan achshent to a random drunkard character a sign of respect to Russian nationals and embracement for diversity? Likely not…
When I draw someone, I don't give much thought to skin color or nationality. Someone is intentionally choosing a stereotype to signal a particular culture, region, or race.
If you asked me to draw a Chinese man, he could be in a suit or traditional or a pair of jeans and t shirt.
Which do I pick? I suppose whatever the scene or situation has called for.
Now, what about a generic "learn Spanish" sentence? What does a Chinese man in traditional wear have to do with the Spanish word for cheese?
Would I find a Chinese man in traditional Chinese clothes speaking Spanish in Mexico? No. Probably not.
What I find is that depictions of stereotypes out of context are an unnecessary distraction pushing a perspective unrelated to the task at hand.
And yes, while it may be minor in contrast to every true and serious topic, if someone introduces themselves at a party as Dr. So and So, I'm curious as to why a title was brought into a peer-to-peer conversation. At best it is a distraction that worst an unnecessary signaling of self.
But you're not going, "this representation doesn't conform to my expectations so it shouldn't exist." It sounds like you're not taking it that far. Am I wrong?
I feel like I understand the situations that make you feel this way, but I'm still left wondering, why? Can you walk me through your emotional reasoning? Which cognitive mechanisms contribute toward the feeling?
Sure. Corporate or HR or someone is pushing for a perspective that is not natural.
Real example: I created documentation for a specific team, showed it to my supervisor, who approved with one requested change: that I embed alternate text for every image. I knew the target team had no sight impaired people so I asked why. Her explanation was that perhaps someone who was sight impaired might at some point want to review the documentation.
Because of a hypothetical future reality, I was instructed to make changes that were not apart of the original requirement, to benefit someone who to all of our known realities does not yet exist.
I was not asked to translate the document into the 30 languages our company employees speak - this is a true reality.
Now we can argue that text can be converted into languages using existing technology and at the time image to text translation does not capture the intent of the flow charts and screenshots that I had created so a human would have to generate these alternate text blobs.
This one person had a perspective that while admirable was not exactly grounded in reality or practicality but was manifested from their own desires, bias, or influence.
Now, my text was not screened for vocabulary levels that match the intended audience so it is possible that I was either being condescending or esoteric but that was not addressed because the concern was brought up. I do believe we each have a desire to shape to shape our world into our vision - and that bias should be met grace, humility, and thoughtful expression when spoken by someone in authority or with power. My frustration is when an unrelated preference or opinion shapes the way a goal is met instead of the completion of the task.
What I'm picking up is that being asked to do work that doesn't have a real benefit and being kinda lied to about it is the issue. Ya?
Help me tie it back to being a viewer or consumer of media. I'm not finding a clear path back to the original topic of seeing representation and feeling a certain way
It's about trends substituting for actual empathy.
Accessibility for sight impaired is hot in tech the last couple years, so parent poster had to work it in to be "inclusive" despite knowing for a fact that zero blind people use the software.
This gets extra spicy when you start comparing "ethnic groups we are supposed to care about a lot" with the actual, very diverse actually, ethnic fingerprint in tech.
> Would I find a Chinese man in traditional Chinese clothes speaking Spanish in Mexico? No. Probably not.
You'll certainly find Chinese men in China learning Spanish using Duolingo ahead of a trip to Mexico. Should they only look at traditional depictions of Spanish speaking people while they learn? Or should they, the user of the app, occasionally see someone who they may personally associate with, speaking the language they are currently trying to learn. The learner could literally be a Chinese man wanting to speak Spanish in Mexico. People of all skin colors and backgrounds regularly learn new languages, and the characters attempt to reflect that.
That's the theory. I'm not sure I get the anger expressed up-thread.
I’m angry because it doesn’t come off like respecting the culture.
Take, for example, me learning Portuguese. I’d like the illustrations to look like people who live in Portugal, obviously. Same for if I’m learning Nepalese, or Swahili.
Not a person of color who is queer and lives in LA. That’s fine for the English course. Not the Swahili course.
Do you believe that any of those countries are monocultures with exactly one skin tone and sexuality? Or that those languages are spoken in exactly one country?
Honestly, Duolingo is so bad at teaching people a language that I think the language learning community, on the whole, would be better off without the app in business.
Then again, teaching is not really their goal; rather, they'd sell you a language learning game.
I did 3 months of Swahili, and stopped cold turkey when I realized that the gamification part of it kept me coming back, but somehow prevented me from learning anything.
Would have been better to pass my time with Candy Crush.
It prevented you from learning anything? You really would have been better playing Candy Crush?
Like other comments say, there's a lot of misguided hate towards Duolingo. The lack of original criticism is almost a meme.
Duolingo is fun and addicting, and it 100% helps you learn vocab and grammar. You can argue that it's not sufficient to fully learn a language, but no single tool is. You need to mix it with other media, podcasts, music, immersion, and personal lessons. But it's a great starting point. You can argue that there's more efficient, non-gamified ways to learn the basic vocab and grammar, but for a lot of people that's too much effort or not as fun because it's not gamified.
I have learned other languages and I know the feeling of a successful study session. Duolingo is not it. It's a quick hit of, "ok, got the points for the day" or whatever, and the next day same thing. The words I learned went into some place in my brain related to getting points in video games, not into the language part of my brain.
Top comment suggests unionizing against AI. That would be like unionizing against the steam engine during the Industrial Revolution. Unions play (and played) an important role in improving working conditions, safety and pay, but they can’t prevent a sector from using core breakthrough technology without risking destroying the business.
I’m stating the obvious but I think this kind of misdirected opposition to the use of AI will present teething problems during this revolution and may result in the disruption or destruction of many businesses. I think the most interesting and at risk area is Hollywood. Big opportunity if you’re an outsider.
From the comments it sounds like Duolingo is still retaining some contractors to verify AI correctness.
However I’m more concerned with the next 5 years. As AI inevitably gets more integrated in products and services, how will people know when AI is wrong?
My biggest fear is that AI will make us all dumber because people assume the computer is right. I see this already with some end users of the software I write.
I know my programs have bugs, sometimes they are wrong. But the average user believes it to be true because it is on the screen. For some reason that is an authority.
Maybe companies who use AI in their producers should be legally required to retain subject matter experts to verify correctness. Maybe AI should be banned from processing queries about health, or other sensitive fields?
People on pushing side already don’t care about quality of translations. Even Microsoft Windows has lots of parallel movements is creating unnecessarily singular batch of detection, that do make sense when translated back into English.
That’s indeed a bit problematic, I guess we’ll see if they finally feel it.
I've enjoyed Duolingo and paid for a while. One takeaway I have is that it's not really a good way to learn languages. It's too formulaic. People complain that they can't understand actual conversation despite years of consistent duo practice. It is fun and addicting though.
It should be one building block when learning a language. as you said, the gamification works just as well as achievements etc. in video games to keep you going, but add whatever interests you to the mix. like music, tv shows, try to talk to people (not always possible of course), memorize short poems and say them out loud, whatever you can come up with.
I've progressed way faster after I stopped using it. Duolingo is tricking you into thinking you are making progress with all those virtual trophies whereas you aren't making much progress in reality. At some point you just connect to keep the streak going which take a higher priority than the language journey itself ironically.
And then there's a big variation of quality as well, the French and Spanish courses just aren't the same as the Vietnamese one.
It's going to be very weird for HN but I'm learning using Tiktok + a flashcard app + a dictionary. You can probably do the same thing with YouTube Shorts.
There's a lot of high quality and very short films on basic daily subjects and subtitled on this app. It's the perfect material you need as a beginner. At the beginning I was watching one 3 minute video per hour and now I just watch it close to normal speed, progress isn't linear.
Short subtitled videos is the ideal material because you are working on your spoken and written comprehension in the same time.
I use the dictionary app on the new words I don't know about and input them into the flashcard app.
I went from close to nothing to around B1 in one year and half. Whereas with Duolingo I never even reached A2 in three years.
My take with language learning is that there's no magic method, you are going to be tired and consistency beats everything, it's a marathon, not a sprint.
Thanks for that. I never thought of using YouTube Shorts (or even Instagram reels) as a language learning tool -- but you're right, Shorts are always subtitled, and the content is interesting too! (because algorithms)
It's easier to learn when you're actually interested in the content.
Other tools I use:
* https://www.languagereactor.com -- formerly Netflix for Language learning, a Chrome plugin for Netflix and YouTube for multi language subtitles.
* https://youglish.com/ -- search for YouTube videos with subtitles in the language you're learning. For shadowing and practicing pronunciation.
> One takeaway I have is that it's not really a good way to learn languages.
I like the bite sized modules that I do while i am waiting in a long checkout line. Its has one the best inertia resistant UI.
Sure if you are super-motivated there are prbably better ways but with language learning there is a long period of grinding before you start conversing or understanding the language. Most ppl are unable to learn a languange because its really hard to visualize gains and benefits in far future.
As a language learner, there are other activities that are better bang for buck than Duolingo. Duolingo is good for low-effort practice through gamification (which is useful for keep a language alive in your head), but it's not meant to teach a language.
The key to learning a language is comprehensible input, and repetition. Our brains are statistical machines and it naturally retains things that are repeated a lot (and spaced).
Of course there are some frequent and stubborn words/phrases that don't stick, in which case, I put them Anki and SRS them. I don't recommend Ankifying everything -- you'll never have enough time to practice and will get demotivated -- but it's very useful to Ankify stubborn words that don't seem to stick, and practice those only.
For instance, in English there are phrases that all seem alike but have very multiple and different meanings, like "put up", "put out", "put in", "put into", "put down". As English speakers we instinctively know what these mean in context, but for a non-native speaker it's super hard to tell the difference, and the key is just to memorize them in context ("put up the curtains", "put up with Joe's habits", "put up a fight"). Anki is super helpful for this.
I don't know that I have a strategy per se, but just things I've learned by trial and error.
1. Forget Duolingo
2. Pursue comprehensible input in context (there are hundreds of ways to do this, you have to find your own way). Polyglots like Steve Kaufmann and Alexander Arguelles have a few recommendations if you google them.
3. Also, try not to learn words. Learn phrases in context.
4. Ankify phrases that just won't stick.
5. Practice speaking and writing from the get go.
6. Leverage languages you already know to help you learn new languages (e.g. if you already know a Romance language, learning another is easier. Japanese unfortunately is a language isolate, but folks who know Chinese characters are able to pick up the written language faster, and folks who know a language with a similar grammar like Korean are able to pickup the grammatical structures faster -- finding those connections early give you a superstructure to map new ideas to).
Consistency is key, and even the very best language learners take about a year to become reasonably familiar with a new language. Disbelieve anything that promises you fluency in 3 months (even Spanish speakers need more than 3 months to be fluent in Portuguese, a closely related language).
Matt vs Japan on YouTube is probably your best resource. Go watch all of his videos (paying attention to how he shifts over the years) and when you’re done binging, get started! I think you can get through most of his videos in 20-30 hours on 2x. If you’re a native English speaker, that should be doable.
It depends on what one wants out of it. Obviously it's not going to be good as a tool for learning to speak, and it's only ever going to at best be so-so for listening. That said, according to Duolingo I should be right around the cusp of A1 & A2 based on where I am in their program. Based on online reading/writing tests I consistently score right around the cusp of A1 & A2. So it's not total bollocks.
I've been using it less of late though, so progress is slowing. I only use the website and they've consistently been making things worse by mimicking the app. They recently switched to the hearts model that the app has which has had a few effects:
1) I am much more cautious and much more likely to cheat (use the UI to give me a hint). I used to push myself by taking a stab at things I wasn't sure of, and if I failed, so what. Now if I'm not sure I'll get the hint first.
2) I spend less time using it. Between making legit mistakes and stupid errors (tap the wrong key, etc) I burn through my hearts much faster than I'd give up in the earlier setup.
Still, I like it for what it is. It fits more easily into my lifestyle than taking classes or other more active forms of learning. I recognize that this brings limitations. But it is far from useless.
I fully believe it can get you to A1 reading and writing, but it totally falls down for speaking and listening. Which may very well be ok/useful for certain uses, but it's disappointing for people (like me) who mainly wanted to use it to talk to people. I thought I was making a ton of progress that I just wasn't until I started doing something else.
> I fully believe it can get you to A1 reading and writing
For some languages they claim B1 and I believe B2 (Spanish & French?) for reading/writing. I'm only about 1/3 through the spanish course and testing at the A1/A2 range on reading/writing tests, so I'd buy it.
But yes, it's almost useless for speaking. When I've been in conversational situations I can sometimes make out what the person is saying, but I can't talk at all. Which also tracks, at no point does it put your brain in a situation where it needs to come up with & speak arbitrary sentences in response to random events. Different neural pathways.
That's neither good nor bad, it just requires people to have properly set expectation. There is no way to build up conversational skills without having conversations.
This has been my experience as well A1 means nothing if you cannot understand the average speech of a native. Most natives talk extremely fast when compared to the speed of duolingo. Natives also slur words together in a way that duolingo never does.
It’s great that you’ve learned the words to be able to buy groceries or order in a restaurant (basic numbers, yes, no, etc) but if you cannot understand the person asking you questions, then it’s all moot.
By itself it is fun but ineffective. I combined it with twice a week personal tutoring sessions. Now I have someone to explain the grammar and the sentences in Duolingo start to make sense. Start to pick up vocab quicker.
I really love Duolingo and wish it didn't get so much hate.
I wish they’d use AI to translate the English language sentences into different dialects of English. It’s hard enough learning German without having to do it via American English.
Just going automatically down the route of “cut costs” speaks to a lack of imagination.
I studied english in a "traditional" English school back in Italy: study grammar, do homeworks etc. However my listening and speaking was thanks to playing mmorpg (final fantasy xiv). Became friend with a Mexican guy and we talked for hours and hours for many days. When I was trying to immigrate into Canada I had to take an English test (I already took a similar one) and I knew my speaking and listening skills were way below my writing and reading skills. This time after the test my listening and speaking skills were higher than my writing, and all the scores were really high. Very satisfying I must say.
It sounds like they're telling us their product has no value. If AI can reliably perform language translation, then why would anyone bother learning a language? Checkmate, Duolingo.
62 comments
[ 2.4 ms ] story [ 80.6 ms ] threadWe need a better source for this than a random Reddit self post before bringing out the pitchforks.
1. They moved away from community built language courses to ones built by “experts.” To be honest, I think the “expert” ones are much worse. The phrases seem more robotic and the structure… what structure is this?
2. We can discuss diversity another day - but what’s not welcome? Literally every screen having to be some new random combination of race, gender, disability, and cultural background. It’s so forced to the point it feels, for lack of a better word, like an indoctrination. I’m not against diversity - but badly handled stuff like this is how you build resentment. If we’re going to have diversity, why not primarily show people who actually speak the language?
Scattershot diversity is what Hollywood films also are finding out causes anger. It’s when you take all the races, genders, etc., and just make sure there’s at least one of each in every movie. This also ironically cheapens diversity, because everyone doubts whether that character would have been there if it wasn’t for the race card.
Compare this with targeted, meaningful diversity. A film studio might make 4 films: One set in 1500s England, one in modern Congo, one in Brazil, one in Mongolia. The British one is almost entirely white people, the Congo one almost entirely black people, etc.
My point is: Vary up the locale, and keep the diversity accurate to the locale. If the locale is diversified enough, the diverse representation will follow with almost no complaints. It’s not like Europe has a monopoly on stories - I’m sure that every culture has stories untold that could be adapted into interesting material.
Not being pandered to 100% of the time does not mean that someone is being repressed. These are *auxiliary graphics in a phone app*. Not even UI elements. Is this really something you consider worth getting worked up about and resenting?
By your own logic, they are auxiliary graphics in a phone app, so who cares?
On that note: “I have yet to meet anyone legitimately angry for the reasons you claim”
1. Where do you get the authority to determine a legitimate reason? Have you dialogued with them to fully understand, instead of stereotyping?
2. Congratulations, you just met me.
If that's true, how do you approach topics like sex and gender? How does one justify a character being queer, trans, &c? On one had I can see, "their identity should add to the story," but on the other hand in real life plenty of people simply are queer or trans or whatever and completely and have stories to tell that don't relate to their identity.
If that's not true, please disregard. :)
When I draw someone, I don't give much thought to skin color or nationality. Someone is intentionally choosing a stereotype to signal a particular culture, region, or race.
If you asked me to draw a Chinese man, he could be in a suit or traditional or a pair of jeans and t shirt.
Which do I pick? I suppose whatever the scene or situation has called for.
Now, what about a generic "learn Spanish" sentence? What does a Chinese man in traditional wear have to do with the Spanish word for cheese?
Would I find a Chinese man in traditional Chinese clothes speaking Spanish in Mexico? No. Probably not.
What I find is that depictions of stereotypes out of context are an unnecessary distraction pushing a perspective unrelated to the task at hand.
And yes, while it may be minor in contrast to every true and serious topic, if someone introduces themselves at a party as Dr. So and So, I'm curious as to why a title was brought into a peer-to-peer conversation. At best it is a distraction that worst an unnecessary signaling of self.
Sincerely, Sir Pizza Esq., C.E.O. the Third
I feel like I understand the situations that make you feel this way, but I'm still left wondering, why? Can you walk me through your emotional reasoning? Which cognitive mechanisms contribute toward the feeling?
Real example: I created documentation for a specific team, showed it to my supervisor, who approved with one requested change: that I embed alternate text for every image. I knew the target team had no sight impaired people so I asked why. Her explanation was that perhaps someone who was sight impaired might at some point want to review the documentation.
Because of a hypothetical future reality, I was instructed to make changes that were not apart of the original requirement, to benefit someone who to all of our known realities does not yet exist.
I was not asked to translate the document into the 30 languages our company employees speak - this is a true reality.
Now we can argue that text can be converted into languages using existing technology and at the time image to text translation does not capture the intent of the flow charts and screenshots that I had created so a human would have to generate these alternate text blobs.
This one person had a perspective that while admirable was not exactly grounded in reality or practicality but was manifested from their own desires, bias, or influence.
Now, my text was not screened for vocabulary levels that match the intended audience so it is possible that I was either being condescending or esoteric but that was not addressed because the concern was brought up. I do believe we each have a desire to shape to shape our world into our vision - and that bias should be met grace, humility, and thoughtful expression when spoken by someone in authority or with power. My frustration is when an unrelated preference or opinion shapes the way a goal is met instead of the completion of the task.
Help me tie it back to being a viewer or consumer of media. I'm not finding a clear path back to the original topic of seeing representation and feeling a certain way
Accessibility for sight impaired is hot in tech the last couple years, so parent poster had to work it in to be "inclusive" despite knowing for a fact that zero blind people use the software.
This gets extra spicy when you start comparing "ethnic groups we are supposed to care about a lot" with the actual, very diverse actually, ethnic fingerprint in tech.
You'll certainly find Chinese men in China learning Spanish using Duolingo ahead of a trip to Mexico. Should they only look at traditional depictions of Spanish speaking people while they learn? Or should they, the user of the app, occasionally see someone who they may personally associate with, speaking the language they are currently trying to learn. The learner could literally be a Chinese man wanting to speak Spanish in Mexico. People of all skin colors and backgrounds regularly learn new languages, and the characters attempt to reflect that.
That's the theory. I'm not sure I get the anger expressed up-thread.
Take, for example, me learning Portuguese. I’d like the illustrations to look like people who live in Portugal, obviously. Same for if I’m learning Nepalese, or Swahili.
Not a person of color who is queer and lives in LA. That’s fine for the English course. Not the Swahili course.
“Well, not everyone is black. So as long as we have an equal mix of black and white people, we’re fine?”
Obviously, no.
https://blog.duolingo.com/lgbtq-representation-in-duolingo-s...
Also, what an example - are you aware that the VAST majority of Portugese speakers don't live in Portugal?
That's the goal, isn't it? Normalize unicorn genders.
And i fully agree with "If we’re going to have diversity, why not primarily show people who actually speak the language?"
Then again, teaching is not really their goal; rather, they'd sell you a language learning game.
Would have been better to pass my time with Candy Crush.
Like other comments say, there's a lot of misguided hate towards Duolingo. The lack of original criticism is almost a meme.
Duolingo is fun and addicting, and it 100% helps you learn vocab and grammar. You can argue that it's not sufficient to fully learn a language, but no single tool is. You need to mix it with other media, podcasts, music, immersion, and personal lessons. But it's a great starting point. You can argue that there's more efficient, non-gamified ways to learn the basic vocab and grammar, but for a lot of people that's too much effort or not as fun because it's not gamified.
It's competing with TikTok and Candy Crush, not with comprehensive, focused study methods.
I’m stating the obvious but I think this kind of misdirected opposition to the use of AI will present teething problems during this revolution and may result in the disruption or destruction of many businesses. I think the most interesting and at risk area is Hollywood. Big opportunity if you’re an outsider.
However I’m more concerned with the next 5 years. As AI inevitably gets more integrated in products and services, how will people know when AI is wrong?
My biggest fear is that AI will make us all dumber because people assume the computer is right. I see this already with some end users of the software I write.
I know my programs have bugs, sometimes they are wrong. But the average user believes it to be true because it is on the screen. For some reason that is an authority.
Maybe companies who use AI in their producers should be legally required to retain subject matter experts to verify correctness. Maybe AI should be banned from processing queries about health, or other sensitive fields?
if a tree falls in the forest and no one is around to hear it, does it make a sound?
if a language model returns an inaccurate response and no one is left that knows the difference, is the response now valid?
That’s indeed a bit problematic, I guess we’ll see if they finally feel it.
And then there's a big variation of quality as well, the French and Spanish courses just aren't the same as the Vietnamese one.
What did you do instead?
There's a lot of high quality and very short films on basic daily subjects and subtitled on this app. It's the perfect material you need as a beginner. At the beginning I was watching one 3 minute video per hour and now I just watch it close to normal speed, progress isn't linear.
Short subtitled videos is the ideal material because you are working on your spoken and written comprehension in the same time.
I use the dictionary app on the new words I don't know about and input them into the flashcard app.
I went from close to nothing to around B1 in one year and half. Whereas with Duolingo I never even reached A2 in three years.
My take with language learning is that there's no magic method, you are going to be tired and consistency beats everything, it's a marathon, not a sprint.
It's easier to learn when you're actually interested in the content.
Other tools I use:
* https://www.languagereactor.com -- formerly Netflix for Language learning, a Chrome plugin for Netflix and YouTube for multi language subtitles.
* https://youglish.com/ -- search for YouTube videos with subtitles in the language you're learning. For shadowing and practicing pronunciation.
I like the bite sized modules that I do while i am waiting in a long checkout line. Its has one the best inertia resistant UI.
Sure if you are super-motivated there are prbably better ways but with language learning there is a long period of grinding before you start conversing or understanding the language. Most ppl are unable to learn a languange because its really hard to visualize gains and benefits in far future.
The key to learning a language is comprehensible input, and repetition. Our brains are statistical machines and it naturally retains things that are repeated a lot (and spaced).
Of course there are some frequent and stubborn words/phrases that don't stick, in which case, I put them Anki and SRS them. I don't recommend Ankifying everything -- you'll never have enough time to practice and will get demotivated -- but it's very useful to Ankify stubborn words that don't seem to stick, and practice those only.
For instance, in English there are phrases that all seem alike but have very multiple and different meanings, like "put up", "put out", "put in", "put into", "put down". As English speakers we instinctively know what these mean in context, but for a non-native speaker it's super hard to tell the difference, and the key is just to memorize them in context ("put up the curtains", "put up with Joe's habits", "put up a fight"). Anki is super helpful for this.
1. Forget Duolingo
2. Pursue comprehensible input in context (there are hundreds of ways to do this, you have to find your own way). Polyglots like Steve Kaufmann and Alexander Arguelles have a few recommendations if you google them.
3. Also, try not to learn words. Learn phrases in context.
4. Ankify phrases that just won't stick.
5. Practice speaking and writing from the get go.
6. Leverage languages you already know to help you learn new languages (e.g. if you already know a Romance language, learning another is easier. Japanese unfortunately is a language isolate, but folks who know Chinese characters are able to pick up the written language faster, and folks who know a language with a similar grammar like Korean are able to pickup the grammatical structures faster -- finding those connections early give you a superstructure to map new ideas to).
Consistency is key, and even the very best language learners take about a year to become reasonably familiar with a new language. Disbelieve anything that promises you fluency in 3 months (even Spanish speakers need more than 3 months to be fluent in Portuguese, a closely related language).
I've been using it less of late though, so progress is slowing. I only use the website and they've consistently been making things worse by mimicking the app. They recently switched to the hearts model that the app has which has had a few effects:
1) I am much more cautious and much more likely to cheat (use the UI to give me a hint). I used to push myself by taking a stab at things I wasn't sure of, and if I failed, so what. Now if I'm not sure I'll get the hint first. 2) I spend less time using it. Between making legit mistakes and stupid errors (tap the wrong key, etc) I burn through my hearts much faster than I'd give up in the earlier setup.
Still, I like it for what it is. It fits more easily into my lifestyle than taking classes or other more active forms of learning. I recognize that this brings limitations. But it is far from useless.
For some languages they claim B1 and I believe B2 (Spanish & French?) for reading/writing. I'm only about 1/3 through the spanish course and testing at the A1/A2 range on reading/writing tests, so I'd buy it.
But yes, it's almost useless for speaking. When I've been in conversational situations I can sometimes make out what the person is saying, but I can't talk at all. Which also tracks, at no point does it put your brain in a situation where it needs to come up with & speak arbitrary sentences in response to random events. Different neural pathways.
That's neither good nor bad, it just requires people to have properly set expectation. There is no way to build up conversational skills without having conversations.
It’s great that you’ve learned the words to be able to buy groceries or order in a restaurant (basic numbers, yes, no, etc) but if you cannot understand the person asking you questions, then it’s all moot.
I really love Duolingo and wish it didn't get so much hate.
Just going automatically down the route of “cut costs” speaks to a lack of imagination.
Are you going to tell me that the option for British English has been there all along, and I just couldn’t find it?