The problem is that the tools you need are "listening to and interacting with people in the target language". Human language is a problem of interpreting the creative social signals of other human beings, and there's no way to automate that. With the Web, YouTube and internet radio, the actual amount of target-language material that a person can expose themselves to is hugely, hugely more available than it was 20-30 years ago.
You have to keep in mind what a tool is trying to accomplish relative to the personal development it takes to grow into another human language. Different levels of learners require different kinds of instruction. Bilingual instruction is quickest if a person has zero knowledge of the language whatever. Absolute beginners need direct instruction in phonology. An Anki deck is to build basic vocabulary or literacy for a person who has some. Duolingo is there to get you comfortable with basic grammar or phrases, to get you interacting with native speakers in the first place once you stumble off the plane. A more advanced speaker might need accent reduction work.
Anki is fantastic at flashcards. Duolingo is fantastic at translation exercises, way better than what we had a few decades ago. Definitely the author in the OP should be using Anki rather than postcards on her bed.
"Mastery"-level linguistic tasks include things like writing creative poetry in the language that another person finds moving, describing the movements of a complex machine, or composing an essay in a specific literary style—things that native speakers may find difficult without formal training. There's just no way Duolingo can do that.
I took a survey recently that clarified a lot of this for me. The questions were all things like: What percentage of your time do you spend listening to music in the target language (and your native language)? What percentage of your time do you spend watching moveis in the target language (and your native language)? What percentage of your time do you spend reading books in the target language (and your native language)? What percentage of your time do you spend thinking and interacting with other people and thinking in the target language (and your native language)? If I spend 1% of my time or less in the target language and 99% in my native language, it's no wonder that I have plateaued in it.
The fact that digital tools can't be used to achieve total native-level fluency shouldn't be used as a reason not to try to improve what's available. It should be possible at least to use digital tools to learn to listen to and understand foreign language media at a useful level... I'd love to be able to watch a movie in the language I'm studying and understand 90% of what's going on. But that seems like an impossible dream given the available tools and my current rate of progress.
If I could reach just that goal, I'd be delighted to explore other more personally immersive avenues for taking my language skills to the next level.
If you go to language-learning forums (e.g. the Japanese learning subreddit) you’ll find a lot of people who spend 95% of their time wanking with learning tools, and 5% (or less) actually studying. Everyone thinks the tools are the problem. Moreover, fixing tools is easier and less painful than doing the grueling work of learning a language. As a result, lots of people get caught in the trap of believing that fluency is One Good Tool Away. It isn’t.
The most important part of the parent’s comment is the one you overlooked: you need to listen and interact with people in the native language. The more hours you spend talking with people in your target language, the better you’ll become. Tools help on the margins, but there’s no magic bullet that will eliminate the time and exposure required.
Unless you're starting as a small child, you will never pick up words that do not occur frequently just from listening and interacting. Or it will take many more years than it ought to get to a given level.
Knowledge of a large number of these words is required to become competent, because, as a category, those words are coming at you all the time. Problem is, you don't know which one of the many thousands is coming your way next on any given day. The author of this article makes the same remark, basically:
> But there’s also an enormous amount of low-frequency words and syntax that even native speakers might encounter only once a year. Knowing any one of these “occasional” words or phrasings isn’t essential. But in every context — a book, an article or conversation — there will probably be several. They’re part of what gives native speech its richness.
If you don't know these words, you don't know a lot of what you're listening to. You might have this explained to you, and so then in that session you are okay with those words. However, without follow-up repetition, those words will soon evaporate.
I live in a highly multicultural country, Canada. Here you can encounter immigrants who have been here 25 years or more, whose English still sucks; and it's not always due to only associating with speakers of their native language. They use English everyday and interact with English speakers, all right. It's simply due to not mustering the academic wherewithal to study properly.
>you’ll find a lot of people who spend 95% of their time futzing with learning tools, and 5% (or less) actually studying
Exactly my point. It shouldn't be necessary to spend all that time futzing with tools, the tools should be there, ready to use.
>you need to listen and interact with people in the native language.
Such people are not available in all areas. Online availability does not scale. These are precisely the issues that digital tools should be good at addressing, but have failed to do so.
”Exactly my point. It shouldn't be necessary to spend all that time futzing with tools, the tools should be there, ready to use.”
It isn’t necessary. The tools are fine, and/or improving them won’t solve the fundamental problem. People are just procrastinating.
”Online availability does not scale. These are precisely the issues that digital tools should be good at addressing, but have failed to do so.”
Short of making an AGI that fluently speaks your target language, there are no obvious improvements to learning tools that will address the fundamental problem: you need to talk to actual humans.
> there are no obvious improvements to learning tools that will address the fundamental problem: you need to talk to actual humans
When learning to speak and listen, you need to already understand perhaps 80% of what actual humans are speaking in order to learn the other 20% you don't already understand. If you speak to someone and only understand say 30% of what they're speaking and they you, you won't pick up any of the 70% you don't understand, assuming the native speaker even wants to continue talking to you. Interactive software tools must deduce what vocab and grammar you can already understand and speak only that plus the 10% extra it wants you to practise and reinforce. And those tools don't exist. Good language teachers who can do the same are expensive.
Reading materials are far better in this regard, but even there, most of them use a specific learning sequence as defined by national language testing and don't cater for most learners who learnt haphazardly and thus whose current knowledge is scattered all over that continuum.
If that is true, then not only digital tools, but all books, classes, tests and every other technique aside from what you recommend for language learning are fraudulent wastes of time.
Language learning courses and tools are just ways to kickstart normal language acquisition, so you don't have to go through the years of gurgling like a baby has to. Learning to communicate in a new language isn't exactly like learning some other kind of skill--it's much, much huger.
When you're speaking your native language, you're decoding and processing complex social signals in real time at multiple levels of structure (phonological, morphological, syntactic, semantic, pragmatic). Even the best AIs can only get a couple of these levels of structure, with significant error rates, and only in a handful of well-studied languages (there are hundreds or thousands of natural languages). How much time does it take for you to learn 1000 vocabulary words? A natural language will have tens of thousands. A comprehensive grammar of the English language would be dozens of heady volumes and have less information about English grammar than any competent native speaker.
Learning a language is just an astonishingly huge task. No course or system can cover it all. Mostly they're just trying to make things easier for you.
I have an uncle who speaks 6 different languages; he really likes picking up new ones. I once had a discussion about how he could pick them up so quickly and he basically said what the parent was saying: use it, practice it, read newspapers in it, speak to people that speak it too. Immerse yourself, accept that you'll be uncomfortable for a while, and before long, you'll be able to hold a simple conversation.
He also said you know you are fluent the day you can casually joke in real time with people and make them laugh (at the joke!)
Oh, absolutely. But I think they're going to be more specialized tools to work on different skills.
For example, I used to use the perapera-kun plugin and I guess yomichan does a similar thing, but that sort of tool—but say, for closed captions—could help listeners get more comprehensible input from movies. Like, here's one I have to try. https://chrome.google.com/webstore/detail/language-learning-...
Or just apps or social networks that help connect speakers in other languages. I know of a Discord server, that lets me practice a target language; the group just meets at an inconvenient time for me.
> I'd love to be able to watch a movie in the language I'm studying and understand 90% of what's going on. But that seems like an impossible dream given the available tools and my current rate of progress.
You could try making your own tools. Export a vocabulary list from the flashcard program you're using. Download subtitles for a bunch of movies. Cross-reference them with the vocabulary list to find the movie you'll be able to understand the best.
That's the basic idea behind a tool I'm building for myself, but for finding relevant example sentences on https://tatoeba.org instead of movies.
In my personal experience, the easiest way to learn a second language is to immerse yourself in a particular country/culture and interact with the locals, get a job, etc. Live there for at least a year. It worked for me, I learned English and it worked for my wife, she learned French.
Are you French and your wife English-speaking? That’s what the OP article mentioned as “école horizontale”. The author lived in France for years but her family spoke English at home and thus her French proficiency never got great.
In my experience living abroad, the fact that so many people speak pretty good English is both a blessing and a curse. We could have a conversation in French but it’d be pretty boring. So my friends and I always speak in English but I don’t get better at French.
To learn conversational language, you really do need to converse. It is a much faster route to basic fluency. Apps like Marco Polo or even Zoom or Skype along with a native speaking contact are going to work more quickly than any gamified experience where you're only taking to an app.
It is a wonderful privilege to have the money and people available to pursue those options, but it is a privilege many do not enjoy.
Meanwhile, for many of those many, learning a new language is more than a fun pastime or intellectual challenge, it is a matter of economic or even physical survival. Duolingo's mascot likes to breezily inform me that most people using Duolingo in Sweden are refugees trying to learn Swedish. Don't those people deserve more emphasis on practical mastery, and less on social engagement? Do they have the money and opportunity to pursue quality interactions with native speakers?
If digital tools could provide reasonable sub-levels of mastery short of full conversational fluency, like ability to read a newspaper, understand a news broadcast or sitcom, babysit a child, or hold a business conversation in an office, I wouldn't complain. But I'm not aware of any digital teaching tools that prepare a person adequately for any of that.
Fine, digital tools are inadequate for reaching full conversational fluency. But should we then let them off the hook for not delivering any level of practical skill in a reasonable amount of time?
Is it really impossible for them to do any better at delivering real-world value?
As an adult learner, you will never get far simply by listening and interacting. You need listening and interacting, plus intensive vocab memorization and a good modicum of grammar study.
In some ten years of being married to my Japanese wife, speaking as much as possible, and consuming Japanese entertainment (movies, music), I got pretty good at what you might call "household Japanese". That's it. I would forever be stuck at that stage if I didn't embark on intensive study.
You have to keep in mind what a tool is trying to accomplish relative to the personal development it takes to grow into another human language. Different levels of learners require different kinds of instruction. Bilingual instruction is quickest if a person has zero knowledge of the language whatever. Absolute beginners need direct instruction in phonology. An Anki deck is to build basic vocabulary or literacy for a person who has some. Duolingo is there to get you comfortable with basic grammar or phrases, to get you interacting with native speakers in the first place once you stumble off the plane. A more advanced speaker might need accent reduction work.
Anki is fantastic at flashcards. Duolingo is fantastic at translation exercises, way better than what we had a few decades ago. Definitely the author in the OP should be using Anki rather than postcards on her bed.
"Mastery"-level linguistic tasks include things like writing creative poetry in the language that another person finds moving, describing the movements of a complex machine, or composing an essay in a specific literary style—things that native speakers may find difficult without formal training. There's just no way Duolingo can do that.
I took a survey recently that clarified a lot of this for me. The questions were all things like: What percentage of your time do you spend listening to music in the target language (and your native language)? What percentage of your time do you spend watching moveis in the target language (and your native language)? What percentage of your time do you spend reading books in the target language (and your native language)? What percentage of your time do you spend thinking and interacting with other people and thinking in the target language (and your native language)? If I spend 1% of my time or less in the target language and 99% in my native language, it's no wonder that I have plateaued in it.