What's the best way to learn a new language?
bbc.com
bbc.com
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
The article speaks of so much research but, if this is indeed the 'answer', uses nothing of it in answering the question. Could have just put that sound byte up top and saved themselves and us the further trouble...
Also, what's feedback even supposed to mean? Like Anki? Like having a speaker correct your grammar from the get-go as you speak, or do you speak a bunch first and learn by stumbling and do they correct major mistakes at first only? There's a million ways to fill meaning into these words. Sustained exposure is obvious, but none of the other words are any guide
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!
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 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.
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 spent 3 months in central america 15 years ago and even _today_ I can converse a little bit in spanish and read spanish reasonably well. There is nothing better than total immersion.
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
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.
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?
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 absolutely don't doubt your experience, but find it interesting that mine has been the exact opposite.
I listen to a lot of German and read a fair amount. As a result, my listening and reading comprehension got pretty good (at least B2). My writing has also improved significantly (probably also around B2). However, I find that this does not transfer well to speaking, which I need to practise separately in order to see a meaningful improvement. After some targeted lessons I'm just about approaching B1.
Perhaps transferability will improve once I reach a certain level of fluency. I think this might have happened when I was learning English. However, this was so long ago that I no longer remember.
For the next language I might try to overemphasise speaking from day one just to see how the learning trajectory differs.
Is this 80% listening, 20% active using a good way to do it?
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.
Could I speak Japanese at that point? No not really… I even had a Japanese spouse! But we spoke mostly English at home. I could read quite well, but conversation was very challenging.
Then we moved to Japan. Despite not having a job that requires me to speak Japanese, I got enough live exposure just from chatting with people at the gym or in social activities that now, a few years later, I’ve backfilled all that conversational fluency that was missing. No special extra effort required, just living in an environment where I used the language reasonably often.
Anyways, the point is that all the time spent in Anki laid a rock-solid foundation that merely needed activation in the right environment for active fluency to emerge. Of course I no longer do my daily flashcard drills (and I’ve forgotten how to write quite a few kanji as a result) but the work paid off.
I have learned a lot of languages, and the method that worked best for me was to read books in which I was interested and to watch movies spoken in that language that were interesting, with the periodic help of a grammar and a dictionary.
Traditional language handbooks or other simplified sources have not been as useful as being exposed to a great amount of non-simplified language, which I reread and rewatched until I understood it perfectly.
It was essential for the books that I read and the movies that I watched to be good enough, so that I really wanted to understand what was written or said.
Even in the beginning, I did not use the grammar and the dictionary very frequently, especially with movies, but I attempted to guess the meaning of the unknown words and move forward with the reading or watching, without interruptions. Only later I confirmed or rejected my guesses with the grammar and dictionary.
Were you trying to learn a language or did it just happen to you?
> 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_...
The author, among other topics, makes a case for the creation of personalized decks. That practice has been really valuable to me.
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
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.
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.
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.
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.
For me, learning Spanish required flash cards (https://calcubest.com/languages/spanish/tenses) to actually memorize which conjugation rule applies to which tense. But Language Transfer was definitely what made the rules click.
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.
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.
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'"
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
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.
For me, doing something everyday[Duolingo], and one on one tutorial instruction[1] have had the biggest payoff. The first for rote memorization, vocab, and exposure to grammar. The second for listening, speaking, and filling in grammar gaps. For Americans, the price of international individual instruction is surprisingly inexpensive due to the dollar being the reserve currency. That translates to about €135/month for to learn French. The workload and cost are both sustainable and my weekly session has become one of my most looked forward to things I do each week.
You can learn word "investigation" without context, but not get or set
In fact the term does not appear to exist at all.
Beware free lists on Ankiweb. They are very variable in quality. Frankly better to build your own.
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
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.
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.
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!
Amazing
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.
This references his tapes: https://lukesmith.xyz/articles/learning-european-languages-m...
ChatGPT gave a reasonable looking answer to my prompt "Summarize what is special about how Michel Thomas teaches Language".
Maybe just another case of a highly intelligent person coming up with an "obvious" solution that is great: yet is not quite so obvious to others. He clearly was talented - but also he avoided explaining the rationale behind his method for years.
Get a native speaker, ask them to teach you the very basics, get them to keep speaking the language to you and correct you as you go. It doesn't matter how laborious it is as long as you're _active_.
At the point you can get a basic conversation going you can start actually looking at the grammar and the written language.
As a human being, your brain is made for spoken language first. Writing is still new and cognitively less important.
We can argue what "best" means, but from my experience the fastest is buy a grammar with exercises, read the grammar and do the exercises.
> Get a native speaker (...)
It's extremely hard work for the other person until you're past a certain level. I tried this, and they struggled even though I could already speak a bit, they they were a professional and I was paying them well. If you're at zero, this is fantasy.
- 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)
That makes you appreciate the beauty of the language!
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.
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...
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.
There is NO replacement for starting young.