hckrnws

Did I miss the part where the title was answered?

The article mentions some building blocks like microlearning, explains how researchers test people with, for example, fictional words and shapes to avoid that you draw on prior knowledge, states that "experts make a case for human instruction" (but not which case or how that human instruction should be shaped or structured), and shares shards of how well the author did on the different tests. There's a lot of links, which is nice, so I can dive deeper into the things mentioned (I've read a bit about 'statistical learning' and plan to read the linked paper on microlearning which is new to me), but I am not a step further in what (combination of) method(s) is the "best way" up learn a new language. Did I overlook it or fail to put some pieces together?

Edit: that microlearning paper (10.22034/meb.2022.355659.1066) is a waste of time if you've read the submission whence it was linked and know about spaced repetition. The paper makes a case that society has become more fast-paced since Charles Babbage made the difference engine in the 1800s and so microlearning can help us by breaking down lessons to fit into our day, lowers costs per lesson etc., but might also fragment the learning (and other obvious pros and cons). The most interesting part was a forgetting curve cited from another paper

Things that worked for me:

- sing songs with lyrics and a dictionary at hand (listening and pronunciation)

- get a native speaker girlfriend/boyfriend (slangs and chit chat)

- practice role-playing scenarios like restaurants, movies, cabs, etc. (daily routine)

- finally, read the news and comment with your partner (conversation)

One last tip, always use wikipedia in the language you want to learn (vocabulary)

For me, the sentence method works well.

1. I get new sentences from Glossika (they've thought through which sentences to present, and in what order — i.e., the curriculum). I get a few at a time — between 5 and 50, depending on how difficult the target language is / how close it is to one I already know.

2. I put those sentences into Mochi, with a template that automatically creates and embeds audio files of the target language.

3. I do the learning, memorizing, and reviewing of the sentences in Mochi using FSRS. I practice writing and pronunciation as I go along with the cards. (Using Mochi also helps me maintain languages I've learned in the same place.)

4. I return to Glossika and occasionally cram pronunciation practice from the human-generated audio there (Mochi is TTS, after all).

5. I supplement with TV and radio for immersion. When I reach a higher level, I start reading books.

6. Travel or living abroad, when I can.

The real trick is getting a couple new sentences and using SRS every day. Consistency moves mountains!

I haven't tried, but I think most of the steps you have listed can be done in NotebookLM. Thanks for sharing your workflow. It's great.

https://en.wikipedia.org/wiki/Input_hypothesis

This is basically what Dreaming Spanish does right? Just shit loads of comprehensible input videos.

Curious about any anecdotal evidence about this from people. I have always struggled with languages and have been trying to learn Italian for the past 6 months.

Is this 80% listening, 20% active using a good way to do it?

well anecdotally from studying Japanese for about a year and a half before moving there, it seems right to me, in particular the part about conscious effort not being able to produce spontaneous speech.

I was embarrassed how little I could say after countless hours of flash cards and other methods. I'd literally just comprehend nothing if someone talked to me. But after a few months of just listening it became much easier. I've thrown all the Anki cards away afterwards, it was just a waste of time.

Yes! I stumbled on this idea myself (when trying to learn German) and it works very well. I just read books and listen to audiobooks, starting from a very basic level and then gradually higher level. The talking improves almost automatically, without having to practice it.

Per the same wiki[0], the theory seems like pseudo-science:

> lacks testability, is conceptually ambiguous, and exaggerates the role of “comprehensible input” in language acquisition.

[0]https://en.wikipedia.org/wiki/Input_hypothesis#Critiques_of_...

This has worked for me. Just try to enjoy a self bombardment of the foreign language and hope you will catch on eventually.

15 years in china, with a chinese wife and everyone in my family and my environment speaking chinese with each other did not help me learn more than a few words and phrases of chinese. so just bombardement is not enough, you must be doing some active learning if you want anything to catch on.

I'm in progress learning Vietnamese this way. To me, whether it works or not is no longer a question :)

Were you trying to learn a language or did it just happen to you?

I spent my childhood in a rural town but learning Spanish from various teachers from 4th grade through high school. I always did well but focused too much on the process of Spanish such as getting very good at conjugating verbs without knowing what the meant

After several years away from Spanish I picked it back up in college and began traveling and living off and on in Latin America

I remember the first times I started dreaming in Spanish, or the first time I had a screaming match with someone trying to steal money from me. I would unconsciously think of a phrase in English and constantly be trying to convert it to Spanish all day long. It was the most fluent I’ve ever felt

A few months ago I went on a trip to Central America and was worried my Spanish would have been lost after over a decade away. Turns out that quite a bit is still there

Folks regularly compliment me on my pronunciation(which is hugely important and shows that you’re trying, folks give you so much grace if you don’t know the words but are trying)

I also find that I can speak far better than I can listen. I regularly have to ask people to repeat themselves or slow down, which is frustrating to me but what can you expect after not staying sharp?

Last thing: I’ll echo another commenter who said to listen to music. My high school Spanish teacher had us listening and singing shakira. She’d print off the lyrics and we’d sing along. This was hugely valuable for pronunciation and flow. Also, old Shakira stuff is great

Nothing beats the pressure of using a language all day in a place where they don’t speak your language.

I remember meeting a backpacker from another country who spoke English but would only speak Spanish to when we traveled and would pull out her dictionary regularly and make notes in her notebook. I learned that Germans are crazy disciplined and that that discipline pays off. Her Spanish was amazing after only a few months in the country

> Nothing beats the pressure of using a language all day in a place where they don’t speak your language.

Nothing beats immersion I'd agree. I found self-studying very difficult because sure I could try and read or listen, but I had no one to really judge my writing/speaking responses back. Or you learn how to speak like a textbook written in the 80s.

> I also find that I can speak far better than I can listen.

I had the same problem when traveling with a non-fluent understanding of the local language. It logically makes sense though - you only need to learn 1 way to say a thing, but theres 100s of ways for someone to respond to you.

> Folks regularly compliment me on my pronunciation

Conjugation/grammar & pronunciation go a long way. You can fill in vocabulary gaps by reaching for similar enough words, describing the thing, or offering up the English word for the thing and get there often... provided you can place it within a decently constructed sentence.

I also find knowing the local way of saying umm/uhh helps a lot so people understand you are slowing down/thinking/struggling for the right words.

There’s a hybrid approach that works pretty well now in the remote era.

I’ve used a platform called Baselang, which basically gives you unlimited access on demand to get in zoom with people in Latin American countries to have conversations in Spanish. They do have a structured curriculum but actually having direct 1:1 conversations is not too far from actually being in country and practicing.

I have no connection to the service except as a customer and there may be others as well. It’s a model I recommend. I’m already fluent in Spanish but it gets pretty rusty and my vocabulary fades so I’ve been using it to stay current.

> I also find that I can speak far better than I can listen.

Interesting. I'm learning Italian while living in Italy at the moment. I'm much better at listening than speaking. I can eavesdrop quite easily. I am still relatively new to the language so maybe there comes a point where it flips?

no, i think its almost universal that you can listen better than you can speak. To speak, you sort of need to be able to express a full thought (even if it has mistakes in it), whereas to listen you generally only really need to get the gist of it.

I think the parent comment was really just about finding it difficult to hear/distinguish words when spoken at a native speed. In which case, sure, you might find it easier to stammer out a few words. But once you get even a basic level of the language, listening is easier.

> I regularly have to ask people to repeat themselves or slow down

I don't know how true it is, but there is a perception that Spanish is often spoken very rapidly by native speakers. I'm sure this is more true of some languages than others, but I noticed it very early on when I attended a bilingual elementary school for a couple years.

The speed of spoken Spanish varies significantly from one place to another.

> I also find that I can speak far better than I can listen.

That's likely because it is far more "controlled by you". You set the pace, marshal your thoughts, and then carefully speak the line.

With listening you have to deal with a lot of components out of your control:

- Speed of delivery

- background noise

- different speakers means subtly different accents

- a "clock" that starts as soon as they await your reply

The best method, by far: https://www.languagetransfer.org/

1) Use Anki with pictures and pronunciation to get necessary vocabulary. But it needs audio to learn pronunciation. Very important. 2) Speak, listen, speak, listen with native speakers in person. _Nothing_ beats this! 3) Evening school is a bonus

I like Anki because it is a calm piece of tech. It has been there for a long time with the same behavior. There is a merit to its boringness. You can also activate FSRS algorithm for supposedly better spaced repetition in profile/deck settings. This was an interesting read: https://github.com/open-spaced-repetition/fsrs4anki/wiki/Spa...

But where do you source the files? I'm willing to pay for a deck, but it seems impossible to find good information without spending a ton of time researching it. Too much noise/signal ratio in this space.

Take a read on this: https://augmentingcognition.com/ltm.html

The author, among other topics, makes a case for the creation of personalized decks. That practice has been really valuable to me.

You're correct this is a problem. I paid for sprachenlernen24.de which is German to many languages and very good. Not sure about other source languages though.

See also https://louisrli.github.io/blog/2023/06/13/effectively-using...

I thought this was about programming languages before I saw it was from BBC, making me ask - what is the best way to learn a new programming language?

I'm guessing the answer is making small things, but what exactly? I've made so many to do list apps I don't know what to do with them

When learning, motivation is first, everything else follows.

At some point I felt the drive to move on from Python as my main language. There was no question of “how”: when I needed or wanted to build anything, I would simply go with Go (later TypeScript) and plow on. It is hard to pinpoint exactly what motivated that drive, but I think it was probably curiosity after seeing examples in other languages, wanting to be more competitive, and—let’s be honest—the basic desire to feel a little cooler in the eyes of peers.

Be mindful of second-order volition here. Like when someone says “I want to quit %BAD_HABIT%”, what they really say is “I want to want to quit %BAD_HABIT”—if they really wanted to quit, they would have already done it. Similarly, if you want to learn a programming language, you are all set (unless it is so esoteric that there are no suitable resources or references, which never happens), but if you want to want to learn a programming language then what you need is some lateral move (tricking yourself, putting yourself in some situation, etc.) that makes you actually want to learn it.

These days learning a new programming language is a more sketchy question, because LLMs drain a few major sources of motivation: you can hardly feel cool for knowing how to program in a new language, because anyone would rightfully assume it was written with an LLM; you increasingly do not actually need to know a language, because a model writes everything for you; the competitive advantage is decreasing. Unlike speaking some human language, there is no society of native speakers that would accept you more or treat you better thanks to you speaking their language.

> I'm guessing the answer is making small things, but what exactly? I've made so many to do list apps I don't know what to do with them

My favorite way has always been to not just build small things, but build small useful things. There is always something that could be better, and there is always a subset of languages best for the task at hand. If it's a CLI, then a language that can compile to binary tends to be best (for me at least), so that already limits the languages somewhat. Then depending on what the task is, it might make sense to learn a new language for it.

Then naturally over the years I've picked up 10-15 languages this way, by just following what each language seems best at, and not being afraid of spending 2-3 weeks writing something basic.

Then for each language you learn, next one gets a lot easier, especially when most mainstream languages today are Algol-like languages and more similar to each other than different.

As any language, the core is "why" do you want to learn it. Is it to add it to a list and that's it? Then you might struggle by creating todo lists or play pretend on Duolingo.

On the other hand, if you do have a goal in mind try to do tiny bits of that.

My goal for natural languages is always connecting with another culture at a deeper level than just using English. If that's the case, you get someone to talk/write to and slowly do it. It won't be instantaneous or dopamine fueled but after a few years you might realise that you've been chatting with someone completely in their language without major hiccups.

For programming languages, I understand that filling a CV is tantalising and useful, so you've got to come up with projects and things you'd actually like to be doing with such a language.

You could say you want to pick up COBOL for a future job, well figure out what would make sense to use it for and go with that.

And if you really cannot think of anything, then you can fall back to make something up: make a game with such a language (even better if it is not meant for games), automate something, recreate a small tool which you find frustrating. And even if after you have read this and still cannot find a thing which gets you, maybe learning this language is not within your current interests and you might start considering to move on.

Create something you actually need, or port something you already created in another language.

I needed a tool to get the contents of a remote zip file without downloading the whole file. I wanted to learn Go, so I created the tool with Go, then I ported it to Rust when I wanted to learn Rust.

Make things that require you to use the parts of the language you don't have a strong grasp on yet, so as to get to know those better. Sorting algos, data structures, a kanren, and a library website would be a good variety pack. And aside from that, reading codebases is also important. Read the code of your CPAN equivalent, your Alexandria equivalent, your Spring equivalent, and your SDL equivalent.

Writing one from scratch gives a lot of understanding to how it works under the hood and in the process you learn right phraseology and treat all languages as computational fronteds.

I recall that my old German teacher taught us that listening to Music in a particular language, and watching TV where they speak a language, were the two best ways to learn a languages.

Her reason for why: Context and various slang words are grasped much quicker compared to the cumbersome process of repeating of words and phrases (She did not omit the need of the latter though).

She was great, 60 years old at the time and had us repeat the lyrics of Rammstein songs in class, her favorite band.

I don't know if music in a new language works so well. Lots of songs have, like, "forced" slang or even changes in pronunciation or syllable stress to meet the constraints of the lyrics. In my country I see lots of people that only listen to music in English but don't have any grasp of it.

Instead, I would go with cartoons or children/preteen's shows first. In adult shows, even when not R-rated, characters usually speak way too fast, or, what is most common, the voices are not mixed very clearly, unlike cartoons.

What worked for me best (for English) was watching Disney movies, the same ones I watched in Spanish.

> She was great, 60 years old at the time and had us repeat the lyrics of Rammstein songs in class, her favorite band

This is hilarious, like "Now, kids, repeat after me, 'te quiero puta'"

> Lots of songs have, like, "forced" slang or even changes in pronunciation or syllable stress to meet the constraints of the lyrics.

I agree. I’d be wary of this as a beginner, but when you get more advanced, it becomes helpful in untangling your hearing from the isolated, intentionally clean and slowed-down setting of a class.

People don’t actually talk like that. Some slur their speech, others have a heavy accent, and others just place emphasis wherever they feel like it. Some kinds of music* work well for giving you an ear for the changes that matter (And even with all the changes, natives still are able to understand most music, so it is a skill to learn).

* I love me some guturals in my music, but it's probably not the best way to train your ear for every day conversation

Regarding the Cartoons totally, way easier to understand them because they speaking slow. I'm learning German right now the same way

Deutschewelle has a daily slow news audio in German in case that is not the source you are using.

> Instead, I would go with cartoons or children/preteen's shows first.

Just be careful. I was watching Dora en Francais, and whenever Monsieur Diego was talking, I was like, woah, I understand it now, but that's cause he was speaking english.

Michel Thomas is the answer (or it was for me anyway as someone previously TERRIBLE at languages)

BBC made a documentary about him where he teaches a French gcse to the 6 worst kids in the school, in I think 2 weeks. https://youtube.com/playlist?list=PL94A517B00A16C187&si=4eAv...

He was also in the French resistance, survived concentration camps and is generally a very interesting person.

https://en.wikipedia.org/wiki/Michel_Thomas

His successor is https://languagetransfer.org, which is just a labour of love by a genius polyglot and language teacher.

So much so, in fact, that the owners of the Michel Thomas IP tried to sue him for stealing the methodology. The EFF, back when they actually did anything, shredded them.

https://arstechnica.com/tech-policy/2018/07/no-you-cant-pate...

Please check Language Transfer out and support him how you can.

I _loved_ Language Transfer when learning Greek. I haven't used it in many years, but at the time remember that I went from just being able to say "Hi" and "How are you" and "Good" to speaking full sentences with my boyfriend at the time in one day. And when visiting Greece later that year I could get by with strangers in everyday interactions easily. It was a mind-blowing learning experience, as someone who is not highly gifted in languages, and I donated to the LT creator.

Now I am learning Swedish. It has been taking me _way_ too long and unfortunately LT doesn't have a Swedish course. Looking at one of these documentaries about Michel Thomas it does indeed look like exactly that kind of approach! And I see he has a Swedish course. I'm excited to give it a try!

> Occupations Nazi hunter

Amazing

The other amazing thing is that all the recordings of him are done with two random people they just dragged in and did on the fly.

They seem so perfectly weighted and with exactly the right increase in difficulty that I assumed they must have been heavily edited / selected.

But basically he kept his methodology secret - literally locked in a safe - because he didn’t trust anyone after his experience in ww2.

It wasn’t until he was really old that someone convinced him to make the recordings. And the tapes are just that.

I learned Spanish just by being immersed and not really worrying about anything.

I mostly just focused on real, practical vocab. And the verb conjugation came with time.

I ignored verb conjugations at first - eg "He eat food."

Then learned present tense and used tricks to speak past and future tense "Tomorrow he eats food" (but you don't even need present tense for that!)

Then learned the simpler of the two ways to speak in the future - it's equivalent to "I am going to __" rather than "I will __" (in Spanish each verb needs conjugation when saying I will, but you use infinitive when saying going to.

Likewise I picked one of the past tenses (one refers to specific point in time, other is just "in the past"). Doesn't matter, in practical usage.

The rest - progressive, imperative, etc all comes with time. You don't really "need" them though. I still don't know the subjunctive tenses (which are sort of hypothetical, feeling etc) and effectively communicate with people about literally anything.

Most important of all, you just have to be humble, get rid of your pride/shame, and be willing and eager to make mistakes. I've spoken with thousands of native speakers and never had a bad experience due to lack of proficiency, even when I knew nothing. This is what most learners of language (or anything) lack, and they therefore are too afraid to ever actually practice. They need a psychologist more than a language teacher.

I’ve gotten so much mileage out of “¿cómo se dice…?”.

for sure. along with just gestures etc... Communicating an idea is generally not all that difficult - you definitely dont need anything even remotely resembling proficiency.

I love brazilian Portugese becuase I love Brazil and it's people and culture. So I listen to a lot of brazilian music and I am always curious about lyrics. I try to sing along, but it's hard sometimes to read it in english and pronouncate, sometimes 't' is 'chi' ... I might be wrong , I am new to the language and I am learning. I have picked up a lot of words in my subconscious and I know what they mean and this is probably a good way learn in my opinion.

In most of Brazil, the "t" or "d" followed by "i" is pronounced as "tch", like in "cheese".

Assimil, L-R method (parallel texts w/audiobook), essentially redoing a grade school education, AJATT, ginormous Anki decks full of sentences...

Does the article actually answer the question of the headline?

No.

the article was padded in history of the author and other nonsense. I asked ChatGPT to summarize only what's in the article. Here's it's summary

TL:DR;

What the article ultimately says is “best”

From the researchers’ perspective:

Early learning is driven by exposure and statistical pattern recognition

Progress requires sustained immersion-like input

True fluency demands long-term interaction, feedback, and social use

Technology helps but does not replace traditional instruction or real communication

There is no endorsement of one magic method. The article’s conclusion is essentially:

Language learning is slow, exposure-driven, cognitively grounded, and requires long-term human interaction.

--- long summary ---

1. We learn languages through statistical pattern detection

The experiment highlights cross-situational learning (CSL) — the brain’s natural ability to:

Track recurring sounds

Notice patterns in how words co-occur

Gradually infer meaning from frequency and context

Do this even without explicit instruction or feedback

The researchers argue this reflects how language learning works in real immersion environments: You are exposed to lots of ambiguous input and your brain extracts structure from repetition.

People can learn very fast by keeping track of statistics in the environment.

So the article emphasizes that language acquisition begins with pattern recognition under exposure, not formal grammar lessons.

2. Microlearning can help — but only at an early stage

Short, repeated sessions (30 minutes per day in the experiment) produced measurable improvement in both Portuguese and Mandarin tone tasks.

However, this was:

Basic vocabulary mapping

Artificial or simplified input

Early-stage acquisition

The article does not claim this leads to fluency.

3. Prior language experience improves pattern extraction

The author performed unusually well in Portuguese partly because:

Knowledge of French and Spanish helped detect grammar patterns.

Familiarity with how languages work improves recognition of structure.

So experience strengthens your ability to exploit statistical learning.

4. Memory capacity and phonological sensitivity matter

The researchers identify core abilities involved in language learning:

Good ear for pronunciation and rhythm

Ability to detect subtle sound differences

Working memory capacity (holding sentences in mind while processing them)

These cognitive factors influence success.

5. Fast fluency claims are unrealistic

The article is explicit:

Achieving fluency requires sustained exposure, interaction, feedback and social use over many months or years.

It references the US Defense Language Institute:

Up to 7 hours per day

~64 weeks to reach basic professional proficiency

So rapid-fluency marketing claims are contradicted by real-world data.

6. Technology is supplementary, not sufficient

Apps, chatbots, VR, and microlearning tools:

Provide additional practice

Improve access

Offer feedback

But they do not replace high-level, deep language study or human interaction.

7. Real proficiency requires human interaction and cultural nuance

The article stresses that:

Knowing words is not the same as understanding what people say back.

A large portion of language is common vocabulary, but real conversation includes rarer words and cultural meaning.

Cultural nuance and idiomatic understanding come from social use.

Live with it, think in it.

Watching kids TV

Learn your own languages' grammar.

Then learn (in all tenses) the below verbs that are (usually) followed by infinitives

Can / am able Must/ to have to To want to

Then, 'to be' and 'to have' (to go with the above).

Vocabulary...including a boatload of infinitives.

i, indeed, had to learn english grammar in order to learn spanish. When they taught us french as kids in school, it was just an exercise in memorization and I had ZERO conception that the various verb tenses actually mapped onto something in english!

This book (English Grammar for Students of Spanish) was useful to me. I assume there's ones for other target languages as well

https://archive.org/details/englishgrammarfo0000spin_q0j9_05...

It’s true that 70% of a language is about ~100-300 words. In linguistics this is called the “core sight set”. If you’re in a pinch traveling I recommend asking an AI for the 300 most frequent word core sight set and cramming these with Anki. You can get gist with about 10 hours of study and be much more useful than 100 hours of Duolingo. With the core sight set and a generous amount of loan words and gesticulation you can communicate practically any necessity to anyone. It will by no means be elegant or poetic but it gets the job done reliably. It’s the 10,000 word long tail of vocabulary where a language shines but it’s the first 300 where it lives and breathes.

I tried this is it didn't work. The most common words are the most versatile too and need context

You can learn word "investigation" without context, but not get or set

I cannot find anything on google nor on google scholar under "core sight set" that has anything to do with language.

In fact the term does not appear to exist at all.

The focus on a small set of core vocabulary is one of the main principles of the Pimsleur method, along with a strict spaced repetition format. When I travel to a new country I always spend about 15-20 hours beforehand doing the 30-minute Pimsleur lessons, just to pick up basic survival vocabulary. I've always been satisfied with the results.

Frequency lists are very useful but learners need context in order to use them, because little word atoms like prepositions, pronouns etc are heavily over represented in the core set. So make sure to study how to use those, and master the core to be/to have verbs too. Some languages have two verbs that roughly translate as to be so you need to crack that too.

Beware free lists on Ankiweb. They are very variable in quality. Frankly better to build your own.

I am constructing my own. Examples are drawn from books, group chats, podcasts, and other media that I consume in the target language (so I know those are either correct, or at worst, as wrong as native speakers are which is good enough for me), shortened as necessary to make it appropriate for Anki

Then I gave my first 100 cards to a native speaker and somehow it's still full of subtle issues and not infrequently also actual mistakes >.<

The two dozen decks I've downloaded aren't always perfect either, but making your own doesn't guarantee it'll be better than another uploader's honest attempt at making a good deck. I do wish Ankiweb was more collaborative though, at least having a bugtracker where people can report mistakes and additions, if not full on code forge functionality. If I'm not mistaken, all you can currently do is leave a review

Just read the language's wikipedia word by word, each time trying to predict the next word. After several repetitions you'll be an expert in that language, easy peasy.