9 ms·

No, Duolingo is just actually bad. They gamified language learning and I have “beaten the game” in a few languages and it did not make me conversational in thos

No, Duolingo is just actually bad. They gamified language learning and I have “beaten the game” in a few languages and it did not make me conversational in those languages. Nor could I read a real world newspaper or Wikipedia page in the target language. That is entirely Duolingo’s fault. The app teaches you facts, but not conversation. The sentences increase in grammar complexity, but not depth or diversity of topics. Even with their language bots, you’re still having super basic conversations about nothing. They could easily fix this by simply improving the quality of the sentences in the later lessons. Or switch to a different testing format, which worked in the beginning but don’t scale well once you’re past the middle and comfortable with the language. Maybe include substantial reading and listening comprehension exercises (i.e., multiple sentences), instead of a single complex sentence. Duolingo has concrete problems with concrete solutions.

yabadabadoes 7y ago
I think the issue is that they want people who use it less than 10 minutes a day to still make progress and feel they are making a lot of progress. Following spaced repetition rules they can only add minimal amounts of vocabulary and they need to spend that vocabulary on intros to new themes that can be completed in 30 minutes or less. Older versions were a little less tuned to keeping the large audience, did typing without seeing the options, and went into some areas of interest of members of each language group. Reverting any of which would be improvements for teaching the most engaged users at the cost of losing 10X as many barely engaged participants. It is the general problem of tuning content to learner's with out an outside authority preventing tuning to learner laziness.
simias 7y ago
Duolingo is useless on its own, especially for harder language pairs. The only reason it's popular is because of the gamification, I find that apps like Memrise or LingQ are massively more useful to actually improve your language skills. Those apps focus on specific skills (vocabulary learning through spaced repetition for Memrise, learning comprehension for LingQ) which means that you have to use other methods to work on other aspects like grammar and conversational skills but at least they attempt to do one thing and they do it mostly well. The only added value I have from using Duolingo is getting some amount of practice actually constructing my target language but even then it's hugely limited because: - You never have any context which makes the experience always completely artificial and makes it harder to get a feel or the language. IMO Duolingo would be 10x better if lessons followed (even very loosely) an internal story instead of just having random, often contrived sentences one after the other. Something as simple as recurring characters, simple common situations etc... It would make it a lot easier to have a mental model of the world you're talking about and it doesn't seem like it would complicate things heavily. - The bulk of the practice involves translation exercises which is limiting and can even be counterproductive. A very important step towards fluency in any language is to stop translating and start constructing the language directly. I'm not a native English speaker but as I write these words I'm not translating a French internal monologue, I know what I want to say and I say it in English directly. It's faster and more natural that way. - The killing blow is that those translation exercises are corrected by, as far as I can tell, some sort of complicated regexp engine. It's very common for a correct-but-unexpected solution to be rejected and when it gets reported it can lingers for years without being fixed. As such you start not to trust the correction given when you don't understand precisely why your sentence was wrong. That last point is really frustrating. I'm a native French speaker and I tried finishing the "French for English speakers" tree using shortcuts to get an idea of the quality and I simply can't reliably unlock the checkpoints on my first try because some of my perfectly fine and idiomatic French sentences end up not fitting the regexp and be marked wrong.
- umanwizard 7y ago
  I'm a non-native French speaker, but I had very similar thoughts: https://news.ycombinator.com/item?id=16827225 https://news.ycombinator.com/item?id=16827225 (I've since been told that people in France do say "le menu", but I stand by the rest of that comment.)

yabadabadoes 7y ago

I think the issue is that they want people who use it less than 10 minutes a day to still make progress and feel they are making a lot of progress. Following spaced repetition rules they can only add minimal amounts of vocabulary and they need to spend that vocabulary on intros to new themes that can be completed in 30 minutes or less. Older versions were a little less tuned to keeping the large audience, did typing without seeing the options, and went into some areas of interest of members of each language group. Reverting any of which would be improvements for teaching the most engaged users at the cost of losing 10X as many barely engaged participants. It is the general problem of tuning content to learner's with out an outside authority preventing tuning to learner laziness.

simias 7y ago

Duolingo is useless on its own, especially for harder language pairs. The only reason it's popular is because of the gamification, I find that apps like Memrise or LingQ are massively more useful to actually improve your language skills. Those apps focus on specific skills (vocabulary learning through spaced repetition for Memrise, learning comprehension for LingQ) which means that you have to use other methods to work on other aspects like grammar and conversational skills but at least they attempt to do one thing and they do it mostly well. The only added value I have from using Duolingo is getting some amount of practice actually constructing my target language but even then it's hugely limited because: - You never have any context which makes the experience always completely artificial and makes it harder to get a feel or the language. IMO Duolingo would be 10x better if lessons followed (even very loosely) an internal story instead of just having random, often contrived sentences one after the other. Something as simple as recurring characters, simple common situations etc... It would make it a lot easier to have a mental model of the world you're talking about and it doesn't seem like it would complicate things heavily. - The bulk of the practice involves translation exercises which is limiting and can even be counterproductive. A very important step towards fluency in any language is to stop translating and start constructing the language directly. I'm not a native English speaker but as I write these words I'm not translating a French internal monologue, I know what I want to say and I say it in English directly. It's faster and more natural that way. - The killing blow is that those translation exercises are corrected by, as far as I can tell, some sort of complicated regexp engine. It's very common for a correct-but-unexpected solution to be rejected and when it gets reported it can lingers for years without being fixed. As such you start not to trust the correction given when you don't understand precisely why your sentence was wrong. That last point is really frustrating. I'm a native French speaker and I tried finishing the "French for English speakers" tree using shortcuts to get an idea of the quality and I simply can't reliably unlock the checkpoints on my first try because some of my perfectly fine and idiomatic French sentences end up not fitting the regexp and be marked wrong.

umanwizard 7y ago
I'm a non-native French speaker, but I had very similar thoughts: https://news.ycombinator.com/item?id=16827225 https://news.ycombinator.com/item?id=16827225 (I've since been told that people in France do say "le menu", but I stand by the rest of that comment.)