For these reasons, I think the most realistic answer for the best way to learn language is private tutoring.
For people who have a specific need to learn a language this is of course true. Or even for people who have an amount of time and effort they are willing to dedicate to that end.
The thing is, though - I’m very unlikely to invest in private tutoring in order to learn a language speculatively. The time and financial requirements are too high a barrier.
But Duolingo is an investment of maybe 10 minutes a day at some moment of downtime. I’d estimate that I’ve spent maybe 200 hours over three years with it, which has taken me from not knowing a language to being conversationally fluent in that language - something I would otherwise just never have done.
Duolingo (or other similar systems) are very obviously not the optimum way to learn a language to fluency. And it’s especially obvious that they wont replace a tutoring service for government and military clients. But a lot of the criticism of them completely ignores the fact that they fill an entirely separate niche which is totally valid.
> I’d estimate that I’ve spent maybe 200 hours over three years with it, which has taken me from not knowing a language to being conversationally fluent in that language
This really jumped out at me. 200 hours to get to conversationally fluent is very, very fast: in the CEFR framework, with proper tuition, this generally gets one to about A2 ("elementary").
How closely related is the language you're studying to the language(s) you already know? Do you supplement Duolingo with other resources? And, last but not least, what's your definition of "conversationally fluent"?
I'd bet that the time spent is being underestimated. Duolingo doesn't track time and works in "lessons" containing a series of something like 15 questions each. You can do one of those in less than 10 minutes, but if you do two, you're over 10 minutes. There's no option to do less than a lesson (even "practice" is arranged as if it were a lesson), and bailing in the middle wouldn't be recorded as anything.
That could well be the case. But I suspect that there's also a great deal of variability in what people mean when they say "conversationally fluent". For example, I completed the entire French tree in Duolingo a few years back, and would consider myself nowhere near "conversationally fluent".
For people who have a specific need to learn a language this is of course true. Or even for people who have an amount of time and effort they are willing to dedicate to that end.
The thing is, though - I’m very unlikely to invest in private tutoring in order to learn a language speculatively. The time and financial requirements are too high a barrier.
But Duolingo is an investment of maybe 10 minutes a day at some moment of downtime. I’d estimate that I’ve spent maybe 200 hours over three years with it, which has taken me from not knowing a language to being conversationally fluent in that language - something I would otherwise just never have done.
Duolingo (or other similar systems) are very obviously not the optimum way to learn a language to fluency. And it’s especially obvious that they wont replace a tutoring service for government and military clients. But a lot of the criticism of them completely ignores the fact that they fill an entirely separate niche which is totally valid.