Seeing "~1000 possible phonemes" casually written like that made me think—how on earth would you even count that? Is 1000 very close? It feels rather high.
There must also be an element of available time for older learners. I started learning a very different language (Croatian) at c.30 and with a demanding job, exercise, work on the house, and a social life it is very hard to fit the lessons and home study in. God knows how it would be with a kid in the mix.
If Scott is correct, does that suggest particular strategies that later language learning should purse to best counteract the effects seemingly observed in the data?
"Aside from saying that learning rate seems high in youth, probably stays high for a while, and then seems to go down in some kind of plausibly-continuous way most marked between 20 and 30, I’m pretty stumped here."
Brains vary widely, and do life circumstances, so it makes sense if there is enormous variation in how and when our language learning skills deteriorate. It would be very surprising if we all lost our language learning ability in a similar way.
It may also be that the knowledge of one's first language actively hinders the acquisition of the second to native levels: calquing might be so effective that one just does that instead. Basically "thinking in English" instead of Spanish or whatever.
I have to point out the misuse of the word "phoneme" at the end here.
The term in linguistics that gets used for raw individual linguistic sounds (considered as discrete things) is "phones". (To the extent that this is meaningful, of course. Some phoneticists would likely insist that we shouldn't be thinking in such discrete terms and that perhaps "phone" is not such a good concept; but I'm going to ignore such concerns.) The thing is, in any given language, lots of sounds are considered equivalent. E.g., the sounds in English that are "r" and "l" would be considered equivalent to a Japanese speaker, while English makes no distinction between aspirated and unaspirated consonants as many languages do. It's these equivalence classes of sounds that are called "phonemes".
Thus, it makes no sense to talk about "the total number of possible phonemes", across languages, because phonemes are not meaningful across languages; the differing equivalences will conflict with one another. (Is voicing significant and aspiration irrelevant? Or perhaps it's aspiration that's significant and voicing that's irrelevant?) What you mean is phones, not phonemes.
I went to the first piano recital I've ever been to of any kind two weekends ago with my niece and nephew ages 7 and 5. There were about 10 other kids of about the same age and I understand most had been playing for 6 months to 3 years.
They were terrible.
The first thing I said to my wife after was that I have a hot take -- I think we start teaching kids piano way too early. They had absolutely zero grasp of what a piano song is supposed to sound like or the rhythm required. I bet a ton of kids get discouraged that after so much practice they still suck and quit at ages 8-10 -- if only they started at 11 or 12 instead, I bet they would be much better and enjoy it much more!
I am quite sure that the average adult with zero piano playing experience would have done much better than these kids with the same amount of practice time.
I would imagine there's a lot more things get in the way of an adult learning a language, compared with a child. The study would probably only be completely useful if it compared adults who could spend all day in an immersive environment, going to school to learn all day etc, just like children do.
As an adult, you generally need a job or source of income to survive. If you are someone migrating to a richer country, you need to speak the local language or you'll be doing a job where nobody needs to speak to you. On the contrary, if you are at a higher-level role then you're probably going to a place where they speak your language/English. (both situations have negative effects on immersion). Maybe, however, you start your own business, in which case you're still only either going to be doing it in a field where your minimal language skills allow you to get by, or you're hiring multilingual speakers to do all the communication and paperwork for you. (your motivation to learn beyond this will probably be low, as running a business can be time-consuming). And of course, no matter where you are or what level of work you're doing...it's still a job that takes up a large part of your day, and if you have a family, you're going to have very little time to purposely learn a language on top of all those responsibilities.
Beyond that, kids will talk to almost anyone who is sitting next to them, whereas as an adult you're going to find it very difficult to make connections unless you can already hold a conversation, so that's another big negative against your exposure and acquisition pace.
My instinct would be there's not a big difference between adults and children when it comes to learning a language. Younger brains may have more neuroplasticity, sure, but their long-term memory is also weaker. It's mostly about the quality and length of daily opportunities, and we're usually more forgiving and patient with a 6 year old learning a language than we are with a 36 year old.
I've always found this theory compelling, though I'm not sure how it squares up with everything else.
The idea is that adults can learn just as well as children, the problem is that they cannot *forget* as well as children.
As a metaphor, imagine you are searching for the fastest route between home and work, but that every time you use any one route, that route become more habitual. At first you might explore quite a bit all the various routes to find which is most efficient. But as you start to identify the best path, you explore less and less because nothing you are finding is quicker than the familiar route. You exploit the efficiencies in the current route, and as a result, it gets more and efficient. Eventually you might get to the point where the old familiar route is so habitual that, even if you discover a quicker route, you can’t quite switch over to the newer route.
Similarly, in the bioeconomy of the brain, your brain is always trying to be more efficient in how it processes information. And when you are learning, your neurons are creating pathways in the brain, and new ways of “wiring together” that are, hopefully, more efficient than the old pathways. But as these pathways get established, the connections get stronger and stronger, and thus more and more efficient. It eventually gets to the point that, even if the brain could potentially be more efficient by going down a newly established pathway, the old route is so familiar/established/efficient/habitual that it cannot quite make the switch to the new route. And so the new route gets neglected, and eventually gets pruned out.
So the reason there is a critical period of learning is not because adults can’t learn (establish new routes), it is because the old knowledge (old routes) are so efficient and habitual that the new knowledge (new route) doesn’t get used and so dies from neglect.
As someone who learned japanese to fluency as an adult, I subscribe to the theory of adults learning faster.
I absolutely learned to read kanji (japanese characters) faster than the average japanese child does. It takes them 12 years of rote learning throughout school to learn the roughly 2000 standard use characters whereas it took me only 3 (up to maybe 6 years if we're being less charitable about what it means to have learned a character). Part of this is because the japanese education system teaches through rote repetition whereas there are cognitive shortcuts you can use such as mnemonics and breaking characters down to their components.
So I think absolutely adults can learn faster in part because we've 'learned how to learn' - it's just that in the majority of cases we either don't have the time to focus, or don't have the benefit of an immersion environment that children do.
I started this by saying two points but it expanded and is now however many points I have.
Accent and idiomatic speaking are different from language lessons. Language teachers give you a little but you can be a fluent speaker but still have an accent or non-idiomatic habits. I know plenty of people who speak perfectly correct English but have accents and non-idiomatic habits. Often they don't consider it too much of an issue because they can be understood and there's nothing technically wrong with it. Sometimes even if there is it's close enough and people will work to understand them because adults often have some reason to interact. Even if it's as simple as making a purchase you'll fight through their weird pronunciation.
Adults and children speak differently and most adults are not willing to step down to more juvenile ways of speaking or to consume more juvenile media. Nor are most adults willing to correct an adult like they might a child. You might be a 40 year old master physician but if you're learning Spanish and you're at an A2 level then the way you should be speaking is the equivalent of a grade schooler. But how many physicians are going to read a Mexican children''s picture book?
Plus the usual things about time spent learning. Also it's worth noting a lot of people stagnate at a relatively low level of language ability and also that many children fail at foreign language acquisition. And I suppose I'll add in that language has a pretty strong cultural component. You notice this especially with ancient languages which have a lot of terms for things that are irrelevant to the modern world and lacks a lot of terms for things that are very important in the modern world. But it's true of other countries/cultures as well. If you exist in a different culture these nuances can be lost on you.
Also I'll throw in the anecdote that my personal experience is that adult language learning is more efficient (more learned per hour) but happens slower due to less hours. Also, it's extremely hard to help illiterate people to read. Much harder than teaching someone who simply learned a different form of literacy. Even if it's a non-alphabetic language like Chinese.
The "Spanish in 36 months" example is a bit unfair given that Spanish is one of the easiest languages for an English speaker to learn. Arabic or Chinese would be much tougher. I've studied Polish for about 34 months. I'm in Poland right now actually in vacation, and still I'm encountering a lot of unfamiliar words. In Spanish (which I studied in high school and achieved pretty good proficiency) most of the complicated words come from Latin and are basically the same as English, not so for Polish!
>4. The world around us is explained entirely by physics.
>Sean Carroll articulates a basic physicist worldview that quantum mechanics and the physics described by the Standard Model essentially explain everything around us.
Sean Carroll cannot even begin to conceptualize consciousness coherently (as evidenced by his discussion with David Chalmers on Carroll's podcast), so explaining it with physics (*at this stage*) seems off the table.
There is one obvious difference between adults and children: correcting adults is rude.
I have small kids. They make all sorts of pronunciation and grammar mistakes. Everybody corrects them all the time, because, that’s what you do.
English is my second language. Nobody ever corrects me. And this makes a huge difference. A two years ago I hired an accent coach over Zoom. I spent 10 hours with him. The amount of progress I made over those 10 hours was tremendous. I improved my pronunciation over those 10 hours more than I did over the last 15 years of daily usage, simply because he was correcting me.
I would greatly appreciate if people would correct my English, but no one does. Even if I request it, I guess it is just awkward to correct adults.
I sometimes tried to provide pronunciation feedback to my colleagues, knowing how much I would value it, but… All I got was kind of awkward reactions.
"The authors wonder if it’s related to people learning languages better while still in school. But this is a high-functioning sample and you would expect many of them to go to college. Also, many of these people are immersion learners, and it’s not obvious why school would be better for immersion learning than whatever comes after (eg the workplace)."
I haven't looked at their analysis and I wouldn't understand the stats if I did, but I would imagine that if the effect were related to being at school, and even if "many of them" do go to college, the change exhibited by those who don't would still show up as a measurable step in the data.
This is spitballing, but at the most hand-wavy level you could come up with a mechanism where, while exposed to intense learning of any sort, the brain adapts in some way to optimise for learning, making all learning during that window more effective. An analogy would be an athlete's cardio fitness improving, which will have a flow on improvement on many physical activities. Once someone moves out of the classroom (or in non classroom-oriented societies, once they have learned most the of things they really need to know) the brain adapts back to optimise for something else - maybe energy usage, or lower stress levels, or something. I seem to recall some data suggesting that old people who maintain intricate hobbies in retirement have better rates of cognition in general, which definitely suggests something adaptive might be going on.
With an M.A. in foreign language teaching, I am humbled and honored by Scott's treatment of this 'silly' statement: "Children learn language more easily" (the first 5 words in that huge study co-authored by Steven Pinker). To make sense, you need to define first a) 'children' (what age!) b) 'learn' (acquiring or what? School twice a week or with parents/peers) c) 'language' (close to first language or not, A1 or C2, age-related vocabulary or perfect grammar or accent) d) 'more easily' (did it "feel hard" or do the results differ). In addition: most native speakers do never reach C2, some just B1.
My conclusion in the nineties was: shrug. What really matters are the practical implications and there I am shocked to see "Children learn language more easily" abused to introduce foreign languages in kindergarten/ primary school. While none of the studies argue for this. Not even high-school has to be the perfect place for it - at least not if done in those sub-par ways they teach languages there. (Good courses in case the language is important for the kids: sure. And for their parents too, I guess.)
Henry Kissinger arrived in the US at age 15 in 1938 with his brother Walter, who was a year or two younger:
"Another overt difference was that Walter shed his Bavarian accent while Henry notably retained his. When asked why this was the case, Walter would tell interviewers, “Because I am the Kissinger who listens.”"
Be interesting to know whether this has any consequences for verbal reasoning in adults. For instance, imagine a smart child with limited access to or little interest in written content up to age 18.
Would that child still have strong verbal reasoning scores on account of them just being smart?
Or would they have average verbal reasoning ability while their other abilities all remained strong?
People interested in this topic will very much enjoy reading Stanislas Deheane's *Reading in the Brain*. Some of the questions we have about language acquisition can be measured with MRI brain scans.
Note, if you buy the book and your copy has black and white copies of the illustrations instead of the original colour plates, download the colour plates here:
Wondering if native fluency would have anything to do with mouth development; if you aren't making those sounds in the growing phase, your mouth won't adjust to accommodate producing them.
Just read the original post by Scott Young. Mostly ok, but "Children regularly become fluent in their home and classroom language, indistinguishable from native speakers; adults rarely do." Sorry?! Around the globe, hundreds of million of kids forced to take foreign language lessons will NEVER 'become fluent in their classroom language", my guess is "less than 10% if it is not English and not required for a job". I have seen them suffer. - And hardly 1 in a thousand will turn out "indistinguishable from native speakers".
Gah, having studied the relationship between memory and language in my PhD, learning Spanish in high school and Chinese after meeting my wife (at 23), I think this is a pretty good summary, but!
1. Phonetic plasticity decreases with language exposure. This is the *primary* reason we think of kids as better at learning languages. A decline here isn't the same thing as "can never learn to say a Chinese/french/Spanish/japanese R (the asshole consonant)"
2. All plasticity decreases to some extent. How much this matters really depends on how committed you are to the language.
3. You need maybe 25% of your *total language exposure* to be in a language to be what most people would think of as fluent. This is simply easier for a baby or child, where fluency also has lower standards but total language exposure is much lower.
4. Linguistic plasticity often means making conpromises between various language features. This doesn't mean you'll be "worse" at your 'native' language, but you'll be measurably less native.
5. It's difficult to measure the 'need' vs. simple percentage of exposure. I highly lean toward the latter theoretically, but functionally they are not distinguishable.
"Recently, I wrote a defense of the uncontroversial-within-cognitive-science-but-widely-disbelieved idea that the mind is a computer."
We've been comparing the brain/mind to the latest Cool Tech since old gods' time. From clockwork to transistors to current tech, and if within ten years we get quantum whatever, it'll be "the brain/mind is an ansible". It's not so much 'widely disbelieved' as 'yet another one, what'll it be tomorrow?' fatigue of "metaphors that enthuse people who like to think that their hammer means all things are nails".
"So Scott says"
For a moment there, I was going "Why is Scott referring to himself in the third person?" 😀
"This isn’t really what anyone is asking, but it will help clarify some later questions: children need to learn some language before age 5 - 10, or they’ll lose the ability to learn languages at all."
Yeah, this is why 'enhancing language' is all part of childcare (for those who think daycares do nothng but change nappies and give bottles so why is it so expensive?):
"The quality of interactions between practitioners and babies and toddlers is important in developing non-verbal and verbal language skills and can impact on achievement later in life. Practitioners play a very important role in this language and communication development.
Babies are born ready to interact with people. They love being spoken to and played with, which helps their brains develop rapidly in the first two years of life. This interaction helps babies learn about language, turn-taking and listening.
Toddlers learn to interact with other children and adults, to express feelings and thoughts, to imagine and problem-solve. In a few short years, children are able to express themselves and use language for thinking, problemsolving, imagining, exploring, and can interact competently in a group of peers.
Below are some practical suggestions on how practitioners can help babies and toddlers to develop non-verbal and verbal skills.
Opportunities for language development occur throughout the day. Talk with babies and toddlers at every opportunity. Routine activities such as getting dressed, nappy-changing and snack-time, provide ideal opportunities for babies and toddlers to hear the same words in the same predictable steps.
Remember, children will learn to understand words before they can say them so give children lots of opportunities to hear new, interesting and varied words. This can be done in a number of ways.
From three to six years, children rapidly increase the number of words they use and change how they use language to communicate and to learn. A child turning three years may have a few hundred words and by the time he/she is six years, may have many thousand words. At that stage, children use complex language to interact, tell stories and describe past events, problemsolve, negotiate, imagine, co-operate and develop relationships. Here are some ideas to enhance language development.
Most importantly at this stage, expand on what children say. Make their sentences longer by adding words or adding another idea. For example, a three-year-old shows a practitioner his knee and comments, 'I have a sore knee'. A practitioner might respond with, 'Oh, you have a sore knee because you fell'.
Open-ended questions stimulate thinking and conversation, as they require more thought and more than a simple one-word answer. Closed questions can stop a conversation and should be used less frequently. Examples of closed questions include questions that test children, for example, 'What colour is that?' or questions that only have a yes/no answer, for example, 'Do you have a dog?'
Use more comments than questions and remember to wait for children to respond. Longer conversations which stimulate children to extend their thinking, are great for their language development."
And in the cases of developmental delay, especially in language acquistion, early intervention and support is vitally necessary or indeed the child will never catch up.
Most children become capable and confident users of language especially when helped by competent adults. However some children may progress at a different rate due to biological or environmental factors. Problems could arise in relation to a child’s capacity to articulate words or to process meaning. This could be due to a language specific impairment or to causes such as hearing loss, neurological impairment or emotional and behavioural difficulties. The term Language Delay is used to describe these cases in general. Children who are difficult to understand or show signs of being at a level of language development appropriate to a much younger child should be referred to a speech and language therapist for assessment."
I see a lot of reports of people claiming that taking MDMA made them much more fluent in a foreign language. I know MDMA floods the brain with serotonin, and children have more serotonin than adults. Could this all be related?
There's a theory about second language learning I find interesting. It states that if you have never learnt a language in adulthood, and wish to learn a language very different from your native language, you're better off first learning a language very similar to your native language, in order to learn how to learn languages as an adult. E.g. a native English speaker would do better learning Spanish and then Mandarin than just Mandarin straight away. I've only seen anecdotal evidence in favour, but the idea seems interesting.
Nomenclatural differences can make discussions in this area hard for people with different backgrounds to productively share. In psychology, there are the explicit (conscious) and implicit (unconscious) learning systems. In neuroscience, these two systems are generally captured by the terms declarative (hippocampal) and procedural (basal ganglia) systems respectively. I’ll refer to the procedural system here in the neuroscientific sense—meaning basal ganglia, (not step-by-step!)
There’s increasing evidence that conscious, working memory systems, which are at the heart of the declarative learning pathway, increasingly come online as children mature. (I wanted to paste a neat graph here after Gathercole & Alloway’s findings, but I couldn't find a way to paste it in--sorry!)
At the same time, there is evidence that the procedural system declines in ability as children mature. This makes a big difference because the procedural system is what captures the complex patterns of your native language and allows you to think and speak intuitively and in some sense, habitually—you don’t even need to think before blurting out what you want to say.
• K. Janacsek and D. Nemeth, "Chapter 2: Procedural memory: The role of competitive neurocognitive networks across development," in The Cognitive Unconscious: The First Half Century, A. Reber and R. Allen Eds.: Oxford University Press, 2022, pp. 22-36.
• F. S. Zwart et al., "Procedural learning across the lifespan: A systematic review with implications for atypical development," J Neuropsychol, vol. 13, no. 2, pp. 149-182, Jun 2019.
Georgetown neuroscientist Michael Ullman’s work is highly relevant here:
• M. T. Ullman, "The declarative/procedural model: A neurobiologically motivated theory of first and second language," in Theories in Second Language Acquisition: An Introduction, B. VanPatten, G. D. Keating, and S. Wulff Eds., 3rd ed.: Routledge, 2020, pp. 128-161.
• K. Morgan-Short and M. T. Ullman, "Declarative and procedural memory in second language learning: Psycholinguistic considerations," in The Routledge Handbook of Second Language Acquisition and Psycholinguistics, 2022.
See also
• D. Pili-Moss et al., "Contributions of declarative and procedural memory to accuracy and automatization during second language practice," Bilingualism: Language and Cognition, vol. 23, no. 3, pp. 639-651, 2020.
• A. M. Graybiel and K. S. Smith, "Good habits, bad habits," Scientific American, vol. 310, no. 6, pp. 38-43, 2014. (This lovely piece on the basal ganglia procedural system also has an excellent explanatory image.)
The bottom line is that some or much of what we see in the differences between how children and how adults learn a language may be explained by the decline with maturity in the procedural system (it’s not all gone, it just takes a lot more repetition to do the trick) coupled with the improvement in declarative system explicit type learning.
Reagrding accents and speed to fluency (C2) the book fluent forever advocates a theory that both are tied to first learning to distinguish sublties in the target language which are otherwise imperceptable in the native language. Such as the different T sounds in Korean to an English speaker, or the different L and R sounds in English to a Korean speaker.
I suspect that children talking to other children and adults in a monolingual environment get pushed so the "good enough" level is higher.
If you're learning a second language as an adult you get to the point where most other adults can understand you, then the urge to keep listening and correcting your accent falls off.
There are some adults who can learn a second language and get to a native level (my father was one with English). But that works only if you are immersed in the second language and it helps a lot to have "perfect pitch" (so you *know* when you're not pronouncing things like those around you and you *care*, even if they don't).
How does this work with accents? That's a "native" level of proficiency, where you have (mostly) the same words, but pronounce them differently. Yet many English learn the RP accent, as it's not native.
Everything I say here is just what I've experienced and heard from random people.
When you learn a second language what you're really doing is making a model of your first language and then constantly chipping away at it so that it resembles the second language. At some point you'll reach "good enough" and stop naturally remodelling the copy into the second language. What counts as "good enough" depends on your needs, and you can find this called the "okay plateau" in certain places. The "okay plateau" is when you've reached the point where you can function for your needs, and any further progress requires deliberate targeting of your deficiencies. So for example, there are many people who learn a language like Latin only so they can read. And then their speech is halting and bad. But considering that language, anybody who gets further than just reading has to do deliberate practice.
In still extant languages, usually most people hit an Okay Plateau somewhere around "completely functional but accented". This happens because the person's first language doesn't have the phonemes, or cadence, or pitch, or stress, of the second language, and so the person can't even hear what they're missing in order to fix it naturally. What they're missing gets filtered out by the brain as noise, similar to how background chatter gets filtered out in a conversation taking place in a crowded room. Your brain is focusing on the model of your first language which it is trying to change into the second language. It can't even begin to hear Chinese Tones from the model, it sees that as background information that doesn't match the first language and so must be unimportant. In Chinese specifically, English speakers tend to have to be taught tones first, or else their Chinese comes out completely incomprehensible.
Scott mentions mistaking L and R, and this happens in Japanese because the Japanese R isn't an L, and it isn't an English R either. It's closer to the Spanish "r" which English speakers can't properly distinguish either, but because Spanish has more R-like sounds, it matters more there. Japanese people practically can't tell if you pronounce a word like 古い (furui) with the wrong R sound. I've never tried it, but I'm curious if they could tell if you just outright replaced it with an L.
All this means that babies learn their first language in a way fundamentally different from an adult's second language. There is no model, there's some innate language acquisition ability that babies have, and then lose at some point. If you pass this point without acquiring a language at all, your brain is going to have to try and construct the language out of nothing, and apparently, it literally can't do this, it can only memorise words. Language is much more than words.
I will say however controversially, nobody learns languages from explicit rules. Learning rules can be a helpful addendum, a kind of way for your brain to "look" at a language "from another angle", but ultimately the process for acquiring a language is some kind of magic that is primarily driven from wanting to learn the language and being exposed to it enough that your brain can built its model from your first. This is why seriously hundreds of millions of people go through language classes in school and come out the other end knowing not the first thing about how to actually speak a language. The only exceptions are those who both have the will to learn, and get the exposure somewhere else (although I suppose theoretically having grammar books read to you is a strange exposure in itself). This happens a lot in Europe and with English for example, as lots of media is in English and isn't subtitled into Croatian or Turkish or what have you. It is entirely possible, though slow and taxing, to learn a language by foregoing any kind of explicit vocabulary or grammar, and just sitting in front of a television all day.
Speed of secondary language ability depends on the languages you have and the language you're going for. Going to Mandarin from English is a much slower process because the two are not very similar. Going from English to German is much faster. But if you already have a second language, it can help learn a third. If you have English and Spanish, French might be faster than if you had just one of them. On the other hand this still depends on the distance between what languages you have and what you're learning. I find it unlikely that knowing English and Arabic will help you learn Japanese much faster. There might be some kind of pliability thing going on though, where having the experience of learning a language in general can make the process easier a second time.
Also if teaching a baby a language early is measurably better than later, can we teach a baby sign language? If the language learning bottleneck is voice, can we use hands?
It's already been mentioned in several comments, but the common thread that seems to be getting overlooked here is that children, as a group, are operating in a reliably different learning context than their adult counterparts. I'm going to summarize the effect as being because, as a kid, it's your job to learn and, the younger you are, the more true this is.
It's just difficult to boil that down to a small set of metrics to measure the effect.
I speak excellent Spanish, mostly acquired as an adult by immersion in a couple of almost English free environments. My pronunciation and syntax will never be mistaken for a native speaker and although I understand grammatical gender perfectly, I cannot consistently make the correct reference. My daughters who grew up in a mixed Spanish-English home have native-language facility in both acquired in exactly the same amount of time. "How quickly" is not exactly the right question.
My girlfriend has been an English tutor for Russians. She's worked with both adults and kids. This experience made her confident that "children learn language faster" is a total myth. Children are stupider, less able to concentrate and pay attention, often forget things that they have just studied. Whatever advantage they have due to their brain placticity, it's completely compensated by everything else.
I think the critical period for phonemic perception is probably more real than this indicates - possibly not that it's absolute, but at least that "learning" strange phonemes is incomparably harder . My experience with linguist geek friends in the con-lang world is that you can be coached to articulate whatever phoneme, but coaching efficacy plateaus when you can't hear the difference between the target phoneme and other nearby phonemes. Experiments that focus on accent (phoneme production) are missing this, and that's reasonable because the overwhelming majority of phonemes in dominant languages (which people are learning in experiment-tractable quantities) are pretty straightforwardly aurally perceptible, but con-langs get into weirder phoneme qualities.
Anyone that tried learning their 2nd or 3rd or Nth language as an adult knows that there is a big hurdle in acquiring a new language that is accepting that you will speak and think like a child for a long time. We feel frustrated, dumb and are tempted to revert to our native language to recover our self-image of a capable adult, thus squandering our acquisition efforts.
I often feel extremely frustrated communicating in English (my 3rd language). Even writing this post is a little painful; I'm much more eloquent in my head, and in my native language, than this!
Maybe because children don't have the same crystalized self-image as adults they will just soldier through the middle stages of a second language acquisition which is when adults often become frustated and lose motivation.
Had to skim the article so I could get to work, but I am a speech language pathologist. As well as making some of the points Scott makes (Adults who have no language exposure don't develop language fully, the deadline for a native speaker accent is puberty) my professors also argued that early language intervention for kiddos with delays and disorders actually saves money by reducing later services that child will need.
I became friendly acquaintances with a woman from Russia who had moved to America with her husband and child. They liked it here, especially the shopping (Costco! She couldn't believe I had never been and insisted on taking me). Not exactly consumerism in the American "house proud" way, though. They lived a bit like college students, and indeed she retained an idealistic peacenik quality, the one thread remaining of "Communist youth". They were very bright - he though not having even owned a car in his life, quickly got into racing, I'm not sure what you call that sport - where you work on the car in your driveway, souping it up, then put it on a trailer and drive to races. He had a successful global business. I had the impression that though they had left with official imprimatur, even for the cat, after awhile it was not strictly advisable for them to return and I think they did so but once.
He never *seemed* to become very comfortable in English, though his taciturnity in my presence made it hard to tell. He may have found Americans (me) banal, and thus not bothered to reveal his comprehension outside of business.
The daughter thrived in school and was almost unimaginably quickly fluent in both language and kid culture.
My friend was interesting, though. Rarely pausing, when she spoke, though that may have been partly frustration borne of the language barrier - our conversations were generally sort of one way. I don't think she ever asked what my husband did for a living, for example, though it actually would have interested her, given what I soon learned were her sympathies. Maybe it was my fault. I *am* a good listener, eager to scavenge other people's lives, and it can be like undamming a stream. She was overflowing with things to say, me trying to unlock the meaning. (The other two would have moved on with an air of, this is boring, Mom, can we go?) What I found fascinating was that she routinely went to the library, from the very first, and checked out books that I would categorize as "difficult" in terms of reading level. Despite being scarcely able to utter a sentence and having learned no English in school. There was a used book store too which she pillaged in a way that suggested books had not been so surplus in Russia. She took home armloads, and read them. Once she came and fed my cat while we were away on vacation, and when I returned I found I had a new copy of "Master and Margarita" because for some reason she knew she didn't like the translation on my shelf.
(I didn't much like M&M, found it a bit of a mess, especially the last third, but keep it as a memento.)
But throughout our years of acquaintance, our conversations never fully evolved from pantomime and guesswork and misapprehension.
Once, for instance, I had casually asked her the name of the town she was from.
I became almost exasperated when despite a torrent of words five minutes later I still didn't have the name of her town.
Five minutes after that, she was chalk-rock drawing a map of the Soviet Union and I was beginning to understand the nature of her parent's work. "So *not* medicine, and the town had no name ...?"
I moved away and miss our " chats". (Though before that he'd rather improbably taught her to drive and bought her a German luxury sedan, and, American style, she ceased walking and biking everywhere.) She mentioned very early on - indicating the houses all shut up, TV screens flashing, that I was the only person in the neighborhood who did what was customary back home in the evening: stroll about the streets at leisure, ready to talk at length with people.
Hi Scott - did anyone contact you regarding the "Language Learning" idea posted on Reddit a few weeks ago? I've been trying it out on Don Quixote, but I'm 1) the world's least qualified person to implement it and 2) very slow at doing anything.
I'd still be interested in hearing your more complete thoughts.
I suppose I'm now in the minority, at least among admirers of Scott, in thinking that "Critical Periods For Languages" include those distinguishing abbreviations such as "et al." and "e.g." from entire words. Omission of the periods is a mental speed bump for me as a reader. On QWERTY keyboards, the period is easy to type.
For whatever it’s worth I’m a 32 year old who has been learning Chinese for almost exactly two years. I study about 20-30 hours a week. It’s remarkable how much easier it is to learn to read than it is to hear and speak.
I have no way of knowing for sure but I guess I’d estimate my CEFR level is around B2. I don’t have any particular reason to doubt I’ll be functionally fluent in the next two or three years at this pace. The Chinese people I speak to on italki and Preply tell me my progress has been pretty fast and I sound fine but honestly I have no way of knowing whether they’re just being kind.
I wish you would’ve spent some time exploring the effect of IQ on language learning. There’s some interesting research on this topic but I think it would be a fun topic to read your thoughts on.
Not sure if someone addressed this, but learning a language after 30 is harder because your *entire life* is more “solidified”, for lack of a better term, in many ways that make it hard to drop everything and immerse yourself. I studied Russian and made little progress until my abroad semester, where my ability improved dramatically. But I was 20, and the important thing here may not have been my brain but being in a place in life where I was both able and expected to drop everything and devote myself fully to language learning. My programs also had tutors paid to talk to me.
After 25 most people don’t have those kinds of opportunities. At 30 I moved to the Netherlands, and noted that the only Americans I observed becoming genuinely fluent in Dutch had lived with Dutch partners, usually in their 20s. At 30 I had already met my (also American) husband. I had a job and faced the general difficulty 30-somethings feel in meeting new friends in new cities. The Dutch 30-somethings typically *had* enough friends and family to make them too busy to spend too much time with me, definitely not the kind of time where they might want to struggle through coaching me in halting Dutch. I paid for lessons, but only got so far.
At 40, with all of the above compounded by children and a home, it’s a wonder anyone even tries to put in the kind of effort needed to get proficient in another language. It might have less to do with our brains and a lot more to do with the shape of our lives.
Getting a native accent is a combination of listening, to yourself and others, and phonological competence. By phonological competence I mean the ability and awareness to make your organs of speech do new things.
I got to native accent in Japanese as an English speaker by recording myself, and a phonology class I took in college gave me the phonological competence to change my accent to sound more native. There are good resources about how to physically make the sounds of different languages, it's fun to stretch and exercise youf organs of speech so you can make them.
"Older research on this topic focused on feral children like Genie, who had been abandoned or abused and so never learned language. They had a hard time learning language even after being reintegrated into society; most never succeeded. But skeptics argued these children had lots of other problems besides lack of language exposure; maybe the abuse and neglect damaged their brains."
With Genie, a complication factor is that she might already have been mentally retarded before her abuse. From Wikipedia: "When she was 14 months old, she came down with a fever and pneumonitis, and her parents took her to a pediatrician who had not previously seen her. The pediatrician said that, although her illness prevented a definitive diagnosis, there was a possibility that she was mentally retarded and that the brain dysfunction kernicterus might be present, further amplifying her father's conclusion that she was severely retarded.[30][10][33]"
In fact, the conclusion that she was mentally retarded was partially why he started abusing and neglecting her.
Genie was also punished severely for trying to speak during her captivity, which might have conditioned her to avoid speaking even afterwards: "Researchers concluded that, if Genie vocalized or made any other noise, her father would beat her with a large plank that he kept in her room."
I think the missing element here is the incentives. Kids and teenagers are incredibly social. Some of these friendships will become life long bonds. They have a very high natural incentive to communicate well with peers, especially if they are new to the country and suddenly immersed in school and other native social life. Adults, however, only have an incentive to learn enough language to get by. Adults in general don't make as many new friends as children. They have an incentive to learn enough to be productive at work, but even that may be within a immigrant community where they can speak their native language part or most of the time. An adult all alone probably could learn as well as children if they had no choice, but in most cases the incentive is much weaker than for children.
"Meanwhile, 36-month-old Spanish children will still be barely saying their first complete sentences. Advantage: adults!"
This... doesn't match up with my experience around 3 year olds? I used to work in childcare and have known many. By that age, I can speak to most kids the same way I would speak to an adult in normal conversation (unless they've been raised on baby-talk...).
Especially if you give them a year to account for the fact their mouths are still forming and compare against 4 year olds, I'd assume higher level of fluency in the child than an adult with 900 hours practice 9/10 times.
It was a pleasure to see me linked in your newsletter this morning, even if it was just to disagree!
I'll admit I was a bit sloppy with the citations in the aforementioned essay. My context for the "adults seem to learn faster, but plateau earlier" comes from Lightbrown and Spada's How Languages are Learned (p. 165):
"In 1978, Catherine Snow and Marian Hoefnagel-Höhle published a study on a group of English speakers who were learning Dutch as a second language while living in the Netherlands. The learners included children as young as three years old as well as older children, adolescents, and adults. On tests administered when learners had been in the country for less than a year, adolescents were by far the most successful learners. They were ahead of everyone on nearly all of the tests. Furthermore, it was the adults, not the children, whose scores were second best. In other words, adolescents and adults learned faster than children in the first few months of exposure to Dutch."
And in Second Language Acquisition Myths, by Steven Brown and Jennifer Larson-Hall, the authors' #1 myth is "Children Learn Languages Quickly and Easily While Adults are Ineffective in Comparison." I take this as evidence that at least a substantial fraction of the SLA expert community believes that adult learning *speed* isn't slower than children, if we hold the environment and method of acquisition constant.
I'm not an SLA researcher myself, but from reading a few survey books on the topic, my sense is that this is a widely held view, if not universal among experts.
The problem I have had with acquiring fluency in other languages is that I think in English. I dream in English. My internal monologue is in English. Thus any other language I learn is halting to speak (though less to read) because it requires a mechanical internal translation process. I suspect those mental processes become calcified with age, even if the actual memorization and grammatical proficiency stay the same or improve. I can learn pronunciation moderately well, but actually speaking at a normal rate is the limiting factor due to my thoughts. I suppose immersion may be the only possible remedy, but I'm a little skeptical even of that. I once asked my cousin, a Frenchman who emigrated to the U.S. in college over 40 years ago (still has a pronounced accent though entirely fluent), whether he thinks and dreams in English or French. He said still mostly French, with just a smattering of English words.
I didn’t start learning a second language till I was in my thirties. By that time 10 years of exposure to mining noise had caused some moderate hearing loss. There were some Russian phonemes that I just couldn’t distinguish. I was terrible at pronunciation but clever enough with the grammar that I managed to hold my own
It’s a pretty major regret that I’ve remained more or less monolingual all my life but I've seen things you people wouldn't believe. Attack ships on fire off the shoulder of Orion. I watched C-beams glitter in the dark near the Tannhäuser Gate…
Oh wait, that was Rutger Hauer in Blade Runner. Never mind.
I do like to hail the usually Hispanic building crews working in my neighborhood with a cheery “Oye como va!” though.
If they mistakenly think I actually speak Spanish I have to pull out my only other canned phrase.
“Ese y lo que estoy diciendo ahora son el único idioma español que conozco.”
If they know enough English, we might go on talk about how much I love Carlos Santana’s cover of the Tito Puente cha cha cha classic.
"This doesn’t seem to match experiment, where the “critical period” for having a perfect accent lasts until age 10 - 12. Also, sometimes talented people who try really hard can have good pronunciation even if they start after that time."
My mother emigrated to the US from Ukraine in 1935 when she was 12 years 11 months old. They went directly to the small town in southeastern Ohio where her uncle lived and owned a couple of prosperous businesses. She went to public school in that town.
When she arrived she did not know any English. She spoke, read, and wrote Russian and could understand and speak Yiddish. She had never been exposed to any language that used the Latin Alphabet.
She was placed in first grade in her public school when she arrived, but graduated high school on time. She went to Ohio State, majored in Chemistry, received a BS, and married my father (a strictly monolingual American who had no interest in or aptitude for foreign languages). They stayed in Ohio where my two brothers and I were borne and raised. She never spoke to us in Russian nor did she attempt to teach us Russian.
In the late 1950s, Ohio State, like other R1 universities began a program in Slavic Studies (Cold War I, people). My mother enrolled in the program and earned a PhD. She thereafter taught Russian language and Russian civilization to undergraduates. Beyond her dissertation she did no academic writing, but she did do public outreach work for the Slavic department via a weekly radio show.
How good was her English? It was almost perfect General American. She could teach college courses and do radio programs in Ohio with out being misunderstood. The only trace of an accent was in the pronunciation of certain words. E.g. the oo in book came out more like boot than like look.
Her brother, who was 6 years younger than she was (6 when he emigrated), had no accent different than the Americans who grew up in the Ohio town where he went to school.
Effort and teaching have a lot to do with it. Hugh Laurie who played Dr. House on TV for 8 seasons with an impeccable American accent, was born in Oxford England, went to university at Cambridge, and played the ultimate upper class English twit Bertie Wooster in the Jeeves TV series. Daniel Craig, who played James Bond in several 007 movies. When he played a detective from the southern US in the first Knives Out movie his accent was unrecognizable. He was better in the 2nd one.
I think in part the advantage of youth may come down to opportunity. My family moved to a Spanish speaking country for 2 years, during which time my 6 year old attended a local, all-Spanish school. By the end of the two years she sounded just like a local. At this point I had been studying Spanish for about 15 years, having started in my 20s, and she now spoke better than me.
Besides her age, the other big difference between us was that she got to spend 7 hours a day, 5 days per week, completely immersed in Spanish. The most I ever got was an hour of class a couple times a week, a few passing conversations during the week, and 3-4 continuous hours on the weekends - I worked in English the entire time.
But during my time living in Spanish speaking countries I met two adults who had started learning in their 20s who sounded just like local speakers (according to locals). One of them went to a Spanish speaking college full time for 5 years, and the other worked in a Spanish speaking office for around 8, so they both had tons of exposure. I think the older you get, the more likely it is your life and livelihood are kind of stuck in your native language, and the harder it gets to absorb many hours of a new language.
Remember Franzisco Pizzaro's first expedition to South America picked up few boys so they would pick up Spanish quickly and serve as interpreters for future second expedition, an important implication of the theory. (Note that Incas failed to use the Spanish they captured, and sacrificed them instead)
> if you try to teach adults and children the same fake language, adults learn faster whether it’s taught implicitly or explicitly.
This seems to be explained than fake languages are designed by adults and natural languages evolve, shaped by our biological instinct, largely in critical period (many grammar or morphology features that a programmer could have invented, wouldn't stick. yes, there's a problem with conlangs that features don't "stick"), so natural languages are easier for children who are in critical period.
I have to imagine there are some studies coming out of places that focus solely on adult language learners and intensive language training, such as the Foreign Service Institute (U.S. State Department) or Defense Language Institute (U.S. Military), that could inform this question some more. These organizations have a mandate to teach languages to professional proficiency in a defined amount of time. There should be some differences in results as well, as I believe the students attending DLI have already been evaluated for language learning ability while students at FSI have not - they are "forced" to learn a language regardless of their evaluated "ability to learn."
Regarding the tension in accent/pronunciation critical windows of months-old vs 12ish-yo, it seems to me like both can be true: it can be that you do start out capable of distinguishing all phonemes, lose "unused" ones within months, but are able to "recover" most of them until a later age.
My own anecdotal experience supports this in that I was unable to distinguish the sounds of b/v until I learnt how to write, but was then able to painstakingly learn this distinction (probably around age 5-6)
Regarding the tension in accent/pronunciation critical windows of months-old vs 12ish-yo, it seems to me like both can be true: it can be that you do start out capable of distinguishing all phonemes, lose "unused" ones within months, but are able to "recover" most of them until a later age.
My own anecdotal experience supports this in that I was unable to distinguish the sounds of b/v until I learnt how to write, but was then able to painstakingly learn this distinction (probably around age 5-6)Q
This is really interesting, bearing in mind that language acquisition isn't the only "critical window" theory out there. Basically any stage theory of the development of a psychological trait will include critical periods. An important one is the Schema theory of intellectual development by Jean Piaget. Schema theory informed Prototype Theory, one of the leading theories regarding how human beings learn and encode information into the Long Term Memory. A schema is a method of actively searching out specific types of information in one's environment, which then forms the basis of what types of categories and concepts one has. Piaget proposed that children begin learning by actively seeking out sensori-motor schema (give a baby an object and they will put it in their mouths, then throw it at the wall). When they pass through this stage they enter the next one, which uses a different set of schemas.
Language, at least to my understanding, is supposed to develop the same way. Babies learn to discriminate between phonemes before they can emit them, they babble using phonemes they hear the adults using around them before they can speak single words, single words before simple two word sentences, etc.
Each of these stages functions much like a critical window, in that an individual must master the earlier stage to some minimum level before they have the skills to even begin the next stage. All this is supposed to be based on brain development, which doesn't occur evenly and incrementally, but in spurts, by different regions.
But whether young children have some sort of advantage over adults in acquiring these sets of skills isn't actually predicted by Piaget's theory, at least to my recollection. Perhaps brain plasticity declines with age?
Children also have an enormous comparative advantage in learning language i think. This is not the question that everyone asks but is probably more important in making life choices.
If you take school as a given, i think its clear doing immersion languages is way more productive than most other things a school might realistically be capable of.
And as an adult there are many useful skills other than language you could spend three years honing that you have the prereqs for but children dont.
Why do so many native English speaker believe that learning languages is important? They don't have to learn any language. The world already speaks theirs.
I'm asking this as someone who speaks English as a second language. Had it been my first, I never would have taken the trouble to learn another.
A very long while ago, when I was studying psycholinguistics and language acquisition, I remember reading that a very young child acquiring their first language is actually learning about the world -- they are learning to categorize the things they see and experience, put them into similar groups, distinguish among non-similar groups and make connections between things. The way they do this is through language: they name things and concepts and the links between things and concepts. This is the related to why children seem to learn language slower than adults. Children are not just learning language; they're learning about the world at the same time. Adults have already learned about the world so learning language is just putting new labels[1] onto things they're already learned about when they acquired their first language.
[1] I don't mean to imply that learning a foreign language is just a matter of memorizing vocabulary. Of course, it's much more complicated that that.
What about the phenomenon where adult learners tend to forget languages as soon as they stop using them (though they come back with use), while they remain constantly available to child learners?
There does seem to be strong evidence for a "critical period" for first language acquisition in early childhood. Cases like Chelsea and Genie suggest a lack of early language exposure leads to permanent deficiencies in grammar and fluency. For second languages, the picture is more complex. Adults and older children can learn quickly in the short term due to their more developed explicit learning abilities. But long-term, older learners plateau at lower proficiency levels.
Children can develop perfect pitch while adults can’t. The best adults can do is learn relative pitch really well. Can’t help but think there is something similar going on with language acquisition/accent as well.
I wonder if there are two effects going on. The first is biological, your brain gets old.
The second is social/exposure. In high school there is a lot of pressure to fit in, and probably not many if any immigrants who speak your native language so you probably end up listening to American music, watching American tv, and spending time talking to American friends.
College is a less communal experience that is typically large enough to have enough foreign students you can hang out with. So you get far less exposure to American culture and language.
One very important factor is how eager your interlocutors are to teach you their language. Obviously parents are extremely motivated here -- you're their child, and they want to communicate with you! But as an adult learner, I found it easier (and more fun) to learn Hungarian, where they were delighted to see me struggling with their language, than German, where they just switched to English at the first sign of difficulty.
Tangentially, it appears that for learning "perfect pitch" (as commonly known), there IS a critical period, and more tantalizingly, that period can be re-opened with certain drugs!
Warning, anecdata incoming: I am a fully proficient (C2) English speaker. I have learned English in primary and high school and I am using it at work daily. My son spent three years (5yo to 8yo) in the USA and he is fully bilingual. I definitely see a qualitative difference between us. For me, speaking English is more effort even when I am doing it fluently. For him there is no difference. For example, I have to explicitly listen to understand lyrics in a song in English, while it just happens in the background for a song in my primary language. For my son, both languages are processed in the background.
This was a great post, but I looked at this guy’s list and I annoyed by #4. “The world around us is explained entirely by physics”. This might be trivially true in the sense that fundamental physics describes the basic stuff of which the world is made up of, but by that logic so does Democritus’s theory of atoms. Tell me how the standard model explains the specific behavior biological organisms (let alone human society) in a way that doesn’t render the word “explain” pretty much meaningless?
Adults are extremely unlikely to get a native accent, and they may plateau around 99.9% on grammar. But they can otherwise get good enough to do university-level work or hold down a white collar job. Adults can also massively outperform children who are raised from birth in bilingual households, but who are otherwise living in a monolingual community.
Most people, child or adult, learn exactly as much as they need to function.
Source: I learned French in my 30s, and I have a certificate from the French government which says I'm tolerably good at it—good enough to go to a French speaking university, at least. I personally know other adults who worked even harder and who are officially qualified to go to law school. After that level or proficiency, the French government stops testing.
One thing that children have a big advantage on is a fully native accent. That window starts closing around 6, and is gone for virtually everyone by 10 or 12. The other thing that children have an advantage on is the last, say, 0.1% of grammatical accuracy. Even extremely good non-native French speakers make a higher level of gender errors than natives. (There's a super neat paper on this but it's in French.)
But if you'd be happy merely being able to take university classes or work a white collar job, then that's a totally achieveable goal. For fun, I've done long pair-programming sessions in French and it's totally doable.
Schooling from age 12-21 is one critical factor. You're required to read, to interact with your peers, and to write. All of those term papers and dorm-room bull sessions make a huge difference. And of course, that many years of schooling will build you a peer group of native speakers.
Where most adults fail is that they build themselves a "native-language bubble" and only leave it when forced to do so. Learning a language to a high level is hard work. Never underestimate the ability of a resourceful adult to avoid doing that work.
My main challenge in my 40s is keeping my academic French alive. I simply don't use it enough. My "household" French is entirely automatic and doesn't really feel much different than my English at this point.
Interestingly, the "bubble" technique works in reverse. You can build a bubble of your target language. And most rapid learning techniques do some variation of this. Once all other possibilities are exhausted, adult brains will finally give in and learn the damn language.
The key question is one of goals: Do you want to be statistically indistinguishable from an educated native adult? Or would you be happy being able to do professional work, socialize easily, and maybe even give an entertaining presentation? All of the latter are achievable, at a price.
Have I invented this, or did I read the headline findings of a study which said that kids who grow up bilingual turn out practically immune to Alzheimer's? Or it may have been Parkinson's.
I knew a family with two kids who were quad lingual. English, Urdu, Farsi and French.
This was a good post, but it should be pointed out that you clipped the quote in a somewhat misleading manner. The full belief in the original post is "Children don’t learn languages faster than adults, but they do reach higher levels of mastery." This neatly sums up the issue and the last part is just as relevant when making educational choices. Far more dubious is point 5, The reason that calorie-based dieting doesn't work isn't because 'your brain has specific neural circuitry designed to avoid starvation and, by extension, any rapid weight-loss'. It's relatively easy for fat people to lose weight fast. What they struggle to do is not to put this weight back on straight away after they stop starving themselves and this is fundamentally a hormone imbalance issue.
I feel like this seriously misinterprets the main claims about the "additional magic language ability" but don't feel the energy to write it up so just register it.
I am related to a man who served in the US army as military intelligence. He got recruited out of college specifically due to his proficiency picking up languages.
He tells me that over his career, he learned 11 different languages, all of which he used to great effect in all his different postings. He also tells me, that he forgot all of them very quickly after he stopped using them.
Meanwhile, I haven’t spoken French except sporadically in 15 years but slip back into it with some very brief rustiness whenever I must speak it. It’s a language I learned in childhood.
Seeing "~1000 possible phonemes" casually written like that made me think—how on earth would you even count that? Is 1000 very close? It feels rather high.
There must also be an element of available time for older learners. I started learning a very different language (Croatian) at c.30 and with a demanding job, exercise, work on the house, and a social life it is very hard to fit the lessons and home study in. God knows how it would be with a kid in the mix.
If Scott is correct, does that suggest particular strategies that later language learning should purse to best counteract the effects seemingly observed in the data?
"Aside from saying that learning rate seems high in youth, probably stays high for a while, and then seems to go down in some kind of plausibly-continuous way most marked between 20 and 30, I’m pretty stumped here."
Brains vary widely, and do life circumstances, so it makes sense if there is enormous variation in how and when our language learning skills deteriorate. It would be very surprising if we all lost our language learning ability in a similar way.
It may also be that the knowledge of one's first language actively hinders the acquisition of the second to native levels: calquing might be so effective that one just does that instead. Basically "thinking in English" instead of Spanish or whatever.
I have to point out the misuse of the word "phoneme" at the end here.
The term in linguistics that gets used for raw individual linguistic sounds (considered as discrete things) is "phones". (To the extent that this is meaningful, of course. Some phoneticists would likely insist that we shouldn't be thinking in such discrete terms and that perhaps "phone" is not such a good concept; but I'm going to ignore such concerns.) The thing is, in any given language, lots of sounds are considered equivalent. E.g., the sounds in English that are "r" and "l" would be considered equivalent to a Japanese speaker, while English makes no distinction between aspirated and unaspirated consonants as many languages do. It's these equivalence classes of sounds that are called "phonemes".
Thus, it makes no sense to talk about "the total number of possible phonemes", across languages, because phonemes are not meaningful across languages; the differing equivalences will conflict with one another. (Is voicing significant and aspiration irrelevant? Or perhaps it's aspiration that's significant and voicing that's irrelevant?) What you mean is phones, not phonemes.
I went to the first piano recital I've ever been to of any kind two weekends ago with my niece and nephew ages 7 and 5. There were about 10 other kids of about the same age and I understand most had been playing for 6 months to 3 years.
They were terrible.
The first thing I said to my wife after was that I have a hot take -- I think we start teaching kids piano way too early. They had absolutely zero grasp of what a piano song is supposed to sound like or the rhythm required. I bet a ton of kids get discouraged that after so much practice they still suck and quit at ages 8-10 -- if only they started at 11 or 12 instead, I bet they would be much better and enjoy it much more!
I am quite sure that the average adult with zero piano playing experience would have done much better than these kids with the same amount of practice time.
>(eg monolingual Asians who cannot distinguish “R” and “L”)
Isn't that mostly Japanese? I know the distinction fully exists in Chinese, and WP says it exists to some extent in Vietnamese and Korean.
I would imagine there's a lot more things get in the way of an adult learning a language, compared with a child. The study would probably only be completely useful if it compared adults who could spend all day in an immersive environment, going to school to learn all day etc, just like children do.
As an adult, you generally need a job or source of income to survive. If you are someone migrating to a richer country, you need to speak the local language or you'll be doing a job where nobody needs to speak to you. On the contrary, if you are at a higher-level role then you're probably going to a place where they speak your language/English. (both situations have negative effects on immersion). Maybe, however, you start your own business, in which case you're still only either going to be doing it in a field where your minimal language skills allow you to get by, or you're hiring multilingual speakers to do all the communication and paperwork for you. (your motivation to learn beyond this will probably be low, as running a business can be time-consuming). And of course, no matter where you are or what level of work you're doing...it's still a job that takes up a large part of your day, and if you have a family, you're going to have very little time to purposely learn a language on top of all those responsibilities.
Beyond that, kids will talk to almost anyone who is sitting next to them, whereas as an adult you're going to find it very difficult to make connections unless you can already hold a conversation, so that's another big negative against your exposure and acquisition pace.
My instinct would be there's not a big difference between adults and children when it comes to learning a language. Younger brains may have more neuroplasticity, sure, but their long-term memory is also weaker. It's mostly about the quality and length of daily opportunities, and we're usually more forgiving and patient with a 6 year old learning a language than we are with a 36 year old.
https://www.businessinsider.com.au/learning-and-aging-2013-1
I've always found this theory compelling, though I'm not sure how it squares up with everything else.
The idea is that adults can learn just as well as children, the problem is that they cannot *forget* as well as children.
As a metaphor, imagine you are searching for the fastest route between home and work, but that every time you use any one route, that route become more habitual. At first you might explore quite a bit all the various routes to find which is most efficient. But as you start to identify the best path, you explore less and less because nothing you are finding is quicker than the familiar route. You exploit the efficiencies in the current route, and as a result, it gets more and efficient. Eventually you might get to the point where the old familiar route is so habitual that, even if you discover a quicker route, you can’t quite switch over to the newer route.
Similarly, in the bioeconomy of the brain, your brain is always trying to be more efficient in how it processes information. And when you are learning, your neurons are creating pathways in the brain, and new ways of “wiring together” that are, hopefully, more efficient than the old pathways. But as these pathways get established, the connections get stronger and stronger, and thus more and more efficient. It eventually gets to the point that, even if the brain could potentially be more efficient by going down a newly established pathway, the old route is so familiar/established/efficient/habitual that it cannot quite make the switch to the new route. And so the new route gets neglected, and eventually gets pruned out.
So the reason there is a critical period of learning is not because adults can’t learn (establish new routes), it is because the old knowledge (old routes) are so efficient and habitual that the new knowledge (new route) doesn’t get used and so dies from neglect.
As someone who learned japanese to fluency as an adult, I subscribe to the theory of adults learning faster.
I absolutely learned to read kanji (japanese characters) faster than the average japanese child does. It takes them 12 years of rote learning throughout school to learn the roughly 2000 standard use characters whereas it took me only 3 (up to maybe 6 years if we're being less charitable about what it means to have learned a character). Part of this is because the japanese education system teaches through rote repetition whereas there are cognitive shortcuts you can use such as mnemonics and breaking characters down to their components.
So I think absolutely adults can learn faster in part because we've 'learned how to learn' - it's just that in the majority of cases we either don't have the time to focus, or don't have the benefit of an immersion environment that children do.
I started this by saying two points but it expanded and is now however many points I have.
Accent and idiomatic speaking are different from language lessons. Language teachers give you a little but you can be a fluent speaker but still have an accent or non-idiomatic habits. I know plenty of people who speak perfectly correct English but have accents and non-idiomatic habits. Often they don't consider it too much of an issue because they can be understood and there's nothing technically wrong with it. Sometimes even if there is it's close enough and people will work to understand them because adults often have some reason to interact. Even if it's as simple as making a purchase you'll fight through their weird pronunciation.
Adults and children speak differently and most adults are not willing to step down to more juvenile ways of speaking or to consume more juvenile media. Nor are most adults willing to correct an adult like they might a child. You might be a 40 year old master physician but if you're learning Spanish and you're at an A2 level then the way you should be speaking is the equivalent of a grade schooler. But how many physicians are going to read a Mexican children''s picture book?
Plus the usual things about time spent learning. Also it's worth noting a lot of people stagnate at a relatively low level of language ability and also that many children fail at foreign language acquisition. And I suppose I'll add in that language has a pretty strong cultural component. You notice this especially with ancient languages which have a lot of terms for things that are irrelevant to the modern world and lacks a lot of terms for things that are very important in the modern world. But it's true of other countries/cultures as well. If you exist in a different culture these nuances can be lost on you.
Also I'll throw in the anecdote that my personal experience is that adult language learning is more efficient (more learned per hour) but happens slower due to less hours. Also, it's extremely hard to help illiterate people to read. Much harder than teaching someone who simply learned a different form of literacy. Even if it's a non-alphabetic language like Chinese.
The "Spanish in 36 months" example is a bit unfair given that Spanish is one of the easiest languages for an English speaker to learn. Arabic or Chinese would be much tougher. I've studied Polish for about 34 months. I'm in Poland right now actually in vacation, and still I'm encountering a lot of unfamiliar words. In Spanish (which I studied in high school and achieved pretty good proficiency) most of the complicated words come from Latin and are basically the same as English, not so for Polish!
Also Polish grammar is really annoying... https://denovo.substack.com/p/making-polish-count
And I've heard Arabic and Chinese are even harder languages. See also: https://2009-2017.state.gov/m/fsi/sls/orgoverview/languages a list of languages by difficulty
>4. The world around us is explained entirely by physics.
>Sean Carroll articulates a basic physicist worldview that quantum mechanics and the physics described by the Standard Model essentially explain everything around us.
Sean Carroll cannot even begin to conceptualize consciousness coherently (as evidenced by his discussion with David Chalmers on Carroll's podcast), so explaining it with physics (*at this stage*) seems off the table.
There is one obvious difference between adults and children: correcting adults is rude.
I have small kids. They make all sorts of pronunciation and grammar mistakes. Everybody corrects them all the time, because, that’s what you do.
English is my second language. Nobody ever corrects me. And this makes a huge difference. A two years ago I hired an accent coach over Zoom. I spent 10 hours with him. The amount of progress I made over those 10 hours was tremendous. I improved my pronunciation over those 10 hours more than I did over the last 15 years of daily usage, simply because he was correcting me.
I would greatly appreciate if people would correct my English, but no one does. Even if I request it, I guess it is just awkward to correct adults.
I sometimes tried to provide pronunciation feedback to my colleagues, knowing how much I would value it, but… All I got was kind of awkward reactions.
So, maybe this might be a little factor?
"The authors wonder if it’s related to people learning languages better while still in school. But this is a high-functioning sample and you would expect many of them to go to college. Also, many of these people are immersion learners, and it’s not obvious why school would be better for immersion learning than whatever comes after (eg the workplace)."
I haven't looked at their analysis and I wouldn't understand the stats if I did, but I would imagine that if the effect were related to being at school, and even if "many of them" do go to college, the change exhibited by those who don't would still show up as a measurable step in the data.
This is spitballing, but at the most hand-wavy level you could come up with a mechanism where, while exposed to intense learning of any sort, the brain adapts in some way to optimise for learning, making all learning during that window more effective. An analogy would be an athlete's cardio fitness improving, which will have a flow on improvement on many physical activities. Once someone moves out of the classroom (or in non classroom-oriented societies, once they have learned most the of things they really need to know) the brain adapts back to optimise for something else - maybe energy usage, or lower stress levels, or something. I seem to recall some data suggesting that old people who maintain intricate hobbies in retirement have better rates of cognition in general, which definitely suggests something adaptive might be going on.
With an M.A. in foreign language teaching, I am humbled and honored by Scott's treatment of this 'silly' statement: "Children learn language more easily" (the first 5 words in that huge study co-authored by Steven Pinker). To make sense, you need to define first a) 'children' (what age!) b) 'learn' (acquiring or what? School twice a week or with parents/peers) c) 'language' (close to first language or not, A1 or C2, age-related vocabulary or perfect grammar or accent) d) 'more easily' (did it "feel hard" or do the results differ). In addition: most native speakers do never reach C2, some just B1.
My conclusion in the nineties was: shrug. What really matters are the practical implications and there I am shocked to see "Children learn language more easily" abused to introduce foreign languages in kindergarten/ primary school. While none of the studies argue for this. Not even high-school has to be the perfect place for it - at least not if done in those sub-par ways they teach languages there. (Good courses in case the language is important for the kids: sure. And for their parents too, I guess.)
Henry Kissinger arrived in the US at age 15 in 1938 with his brother Walter, who was a year or two younger:
"Another overt difference was that Walter shed his Bavarian accent while Henry notably retained his. When asked why this was the case, Walter would tell interviewers, “Because I am the Kissinger who listens.”"
Be interesting to know whether this has any consequences for verbal reasoning in adults. For instance, imagine a smart child with limited access to or little interest in written content up to age 18.
Would that child still have strong verbal reasoning scores on account of them just being smart?
Or would they have average verbal reasoning ability while their other abilities all remained strong?
People interested in this topic will very much enjoy reading Stanislas Deheane's *Reading in the Brain*. Some of the questions we have about language acquisition can be measured with MRI brain scans.
Note, if you buy the book and your copy has black and white copies of the illustrations instead of the original colour plates, download the colour plates here:
https://readinginthebrain.pagesperso-orange.fr/figures.htm
You won't be able to make sense of many of the plates if you only have black and white.
Wondering if native fluency would have anything to do with mouth development; if you aren't making those sounds in the growing phase, your mouth won't adjust to accommodate producing them.
Just read the original post by Scott Young. Mostly ok, but "Children regularly become fluent in their home and classroom language, indistinguishable from native speakers; adults rarely do." Sorry?! Around the globe, hundreds of million of kids forced to take foreign language lessons will NEVER 'become fluent in their classroom language", my guess is "less than 10% if it is not English and not required for a job". I have seen them suffer. - And hardly 1 in a thousand will turn out "indistinguishable from native speakers".
Gah, having studied the relationship between memory and language in my PhD, learning Spanish in high school and Chinese after meeting my wife (at 23), I think this is a pretty good summary, but!
1. Phonetic plasticity decreases with language exposure. This is the *primary* reason we think of kids as better at learning languages. A decline here isn't the same thing as "can never learn to say a Chinese/french/Spanish/japanese R (the asshole consonant)"
2. All plasticity decreases to some extent. How much this matters really depends on how committed you are to the language.
3. You need maybe 25% of your *total language exposure* to be in a language to be what most people would think of as fluent. This is simply easier for a baby or child, where fluency also has lower standards but total language exposure is much lower.
4. Linguistic plasticity often means making conpromises between various language features. This doesn't mean you'll be "worse" at your 'native' language, but you'll be measurably less native.
5. It's difficult to measure the 'need' vs. simple percentage of exposure. I highly lean toward the latter theoretically, but functionally they are not distinguishable.
From the linked piece, insert eye-rolling here:
"Recently, I wrote a defense of the uncontroversial-within-cognitive-science-but-widely-disbelieved idea that the mind is a computer."
We've been comparing the brain/mind to the latest Cool Tech since old gods' time. From clockwork to transistors to current tech, and if within ten years we get quantum whatever, it'll be "the brain/mind is an ansible". It's not so much 'widely disbelieved' as 'yet another one, what'll it be tomorrow?' fatigue of "metaphors that enthuse people who like to think that their hammer means all things are nails".
"So Scott says"
For a moment there, I was going "Why is Scott referring to himself in the third person?" 😀
"This isn’t really what anyone is asking, but it will help clarify some later questions: children need to learn some language before age 5 - 10, or they’ll lose the ability to learn languages at all."
Yeah, this is why 'enhancing language' is all part of childcare (for those who think daycares do nothng but change nappies and give bottles so why is it so expensive?):
https://www.aistearsiolta.ie/en/play/resources-for-sharing/enhancing-language-birth-3-years-.pdf
"The quality of interactions between practitioners and babies and toddlers is important in developing non-verbal and verbal language skills and can impact on achievement later in life. Practitioners play a very important role in this language and communication development.
Babies are born ready to interact with people. They love being spoken to and played with, which helps their brains develop rapidly in the first two years of life. This interaction helps babies learn about language, turn-taking and listening.
Toddlers learn to interact with other children and adults, to express feelings and thoughts, to imagine and problem-solve. In a few short years, children are able to express themselves and use language for thinking, problemsolving, imagining, exploring, and can interact competently in a group of peers.
Below are some practical suggestions on how practitioners can help babies and toddlers to develop non-verbal and verbal skills.
Opportunities for language development occur throughout the day. Talk with babies and toddlers at every opportunity. Routine activities such as getting dressed, nappy-changing and snack-time, provide ideal opportunities for babies and toddlers to hear the same words in the same predictable steps.
Remember, children will learn to understand words before they can say them so give children lots of opportunities to hear new, interesting and varied words. This can be done in a number of ways.
From three to six years, children rapidly increase the number of words they use and change how they use language to communicate and to learn. A child turning three years may have a few hundred words and by the time he/she is six years, may have many thousand words. At that stage, children use complex language to interact, tell stories and describe past events, problemsolve, negotiate, imagine, co-operate and develop relationships. Here are some ideas to enhance language development.
Most importantly at this stage, expand on what children say. Make their sentences longer by adding words or adding another idea. For example, a three-year-old shows a practitioner his knee and comments, 'I have a sore knee'. A practitioner might respond with, 'Oh, you have a sore knee because you fell'.
Open-ended questions stimulate thinking and conversation, as they require more thought and more than a simple one-word answer. Closed questions can stop a conversation and should be used less frequently. Examples of closed questions include questions that test children, for example, 'What colour is that?' or questions that only have a yes/no answer, for example, 'Do you have a dog?'
Use more comments than questions and remember to wait for children to respond. Longer conversations which stimulate children to extend their thinking, are great for their language development."
And in the cases of developmental delay, especially in language acquistion, early intervention and support is vitally necessary or indeed the child will never catch up.
https://arrow.tudublin.ie/cgi/viewcontent.cgi?article=1016&context=aaschsslbk
"Language Delay
Most children become capable and confident users of language especially when helped by competent adults. However some children may progress at a different rate due to biological or environmental factors. Problems could arise in relation to a child’s capacity to articulate words or to process meaning. This could be due to a language specific impairment or to causes such as hearing loss, neurological impairment or emotional and behavioural difficulties. The term Language Delay is used to describe these cases in general. Children who are difficult to understand or show signs of being at a level of language development appropriate to a much younger child should be referred to a speech and language therapist for assessment."
Are Scott Young's books (his blog really wants to give me FIVE OF MY BOOKS FOR FREE) worth reading?
I see a lot of reports of people claiming that taking MDMA made them much more fluent in a foreign language. I know MDMA floods the brain with serotonin, and children have more serotonin than adults. Could this all be related?
There's a theory about second language learning I find interesting. It states that if you have never learnt a language in adulthood, and wish to learn a language very different from your native language, you're better off first learning a language very similar to your native language, in order to learn how to learn languages as an adult. E.g. a native English speaker would do better learning Spanish and then Mandarin than just Mandarin straight away. I've only seen anecdotal evidence in favour, but the idea seems interesting.
Nomenclatural differences can make discussions in this area hard for people with different backgrounds to productively share. In psychology, there are the explicit (conscious) and implicit (unconscious) learning systems. In neuroscience, these two systems are generally captured by the terms declarative (hippocampal) and procedural (basal ganglia) systems respectively. I’ll refer to the procedural system here in the neuroscientific sense—meaning basal ganglia, (not step-by-step!)
There’s increasing evidence that conscious, working memory systems, which are at the heart of the declarative learning pathway, increasingly come online as children mature. (I wanted to paste a neat graph here after Gathercole & Alloway’s findings, but I couldn't find a way to paste it in--sorry!)
At the same time, there is evidence that the procedural system declines in ability as children mature. This makes a big difference because the procedural system is what captures the complex patterns of your native language and allows you to think and speak intuitively and in some sense, habitually—you don’t even need to think before blurting out what you want to say.
• K. Janacsek and D. Nemeth, "Chapter 2: Procedural memory: The role of competitive neurocognitive networks across development," in The Cognitive Unconscious: The First Half Century, A. Reber and R. Allen Eds.: Oxford University Press, 2022, pp. 22-36.
• F. S. Zwart et al., "Procedural learning across the lifespan: A systematic review with implications for atypical development," J Neuropsychol, vol. 13, no. 2, pp. 149-182, Jun 2019.
Georgetown neuroscientist Michael Ullman’s work is highly relevant here:
• M. T. Ullman, "The declarative/procedural model: A neurobiologically motivated theory of first and second language," in Theories in Second Language Acquisition: An Introduction, B. VanPatten, G. D. Keating, and S. Wulff Eds., 3rd ed.: Routledge, 2020, pp. 128-161.
• K. Morgan-Short and M. T. Ullman, "Declarative and procedural memory in second language learning: Psycholinguistic considerations," in The Routledge Handbook of Second Language Acquisition and Psycholinguistics, 2022.
See also
• D. Pili-Moss et al., "Contributions of declarative and procedural memory to accuracy and automatization during second language practice," Bilingualism: Language and Cognition, vol. 23, no. 3, pp. 639-651, 2020.
• A. M. Graybiel and K. S. Smith, "Good habits, bad habits," Scientific American, vol. 310, no. 6, pp. 38-43, 2014. (This lovely piece on the basal ganglia procedural system also has an excellent explanatory image.)
The bottom line is that some or much of what we see in the differences between how children and how adults learn a language may be explained by the decline with maturity in the procedural system (it’s not all gone, it just takes a lot more repetition to do the trick) coupled with the improvement in declarative system explicit type learning.
Reagrding accents and speed to fluency (C2) the book fluent forever advocates a theory that both are tied to first learning to distinguish sublties in the target language which are otherwise imperceptable in the native language. Such as the different T sounds in Korean to an English speaker, or the different L and R sounds in English to a Korean speaker.
Most people will work until it's "good enough".
I suspect that children talking to other children and adults in a monolingual environment get pushed so the "good enough" level is higher.
If you're learning a second language as an adult you get to the point where most other adults can understand you, then the urge to keep listening and correcting your accent falls off.
There are some adults who can learn a second language and get to a native level (my father was one with English). But that works only if you are immersed in the second language and it helps a lot to have "perfect pitch" (so you *know* when you're not pronouncing things like those around you and you *care*, even if they don't).
How does this work with accents? That's a "native" level of proficiency, where you have (mostly) the same words, but pronounce them differently. Yet many English learn the RP accent, as it's not native.
Everything I say here is just what I've experienced and heard from random people.
When you learn a second language what you're really doing is making a model of your first language and then constantly chipping away at it so that it resembles the second language. At some point you'll reach "good enough" and stop naturally remodelling the copy into the second language. What counts as "good enough" depends on your needs, and you can find this called the "okay plateau" in certain places. The "okay plateau" is when you've reached the point where you can function for your needs, and any further progress requires deliberate targeting of your deficiencies. So for example, there are many people who learn a language like Latin only so they can read. And then their speech is halting and bad. But considering that language, anybody who gets further than just reading has to do deliberate practice.
In still extant languages, usually most people hit an Okay Plateau somewhere around "completely functional but accented". This happens because the person's first language doesn't have the phonemes, or cadence, or pitch, or stress, of the second language, and so the person can't even hear what they're missing in order to fix it naturally. What they're missing gets filtered out by the brain as noise, similar to how background chatter gets filtered out in a conversation taking place in a crowded room. Your brain is focusing on the model of your first language which it is trying to change into the second language. It can't even begin to hear Chinese Tones from the model, it sees that as background information that doesn't match the first language and so must be unimportant. In Chinese specifically, English speakers tend to have to be taught tones first, or else their Chinese comes out completely incomprehensible.
Scott mentions mistaking L and R, and this happens in Japanese because the Japanese R isn't an L, and it isn't an English R either. It's closer to the Spanish "r" which English speakers can't properly distinguish either, but because Spanish has more R-like sounds, it matters more there. Japanese people practically can't tell if you pronounce a word like 古い (furui) with the wrong R sound. I've never tried it, but I'm curious if they could tell if you just outright replaced it with an L.
All this means that babies learn their first language in a way fundamentally different from an adult's second language. There is no model, there's some innate language acquisition ability that babies have, and then lose at some point. If you pass this point without acquiring a language at all, your brain is going to have to try and construct the language out of nothing, and apparently, it literally can't do this, it can only memorise words. Language is much more than words.
I will say however controversially, nobody learns languages from explicit rules. Learning rules can be a helpful addendum, a kind of way for your brain to "look" at a language "from another angle", but ultimately the process for acquiring a language is some kind of magic that is primarily driven from wanting to learn the language and being exposed to it enough that your brain can built its model from your first. This is why seriously hundreds of millions of people go through language classes in school and come out the other end knowing not the first thing about how to actually speak a language. The only exceptions are those who both have the will to learn, and get the exposure somewhere else (although I suppose theoretically having grammar books read to you is a strange exposure in itself). This happens a lot in Europe and with English for example, as lots of media is in English and isn't subtitled into Croatian or Turkish or what have you. It is entirely possible, though slow and taxing, to learn a language by foregoing any kind of explicit vocabulary or grammar, and just sitting in front of a television all day.
Speed of secondary language ability depends on the languages you have and the language you're going for. Going to Mandarin from English is a much slower process because the two are not very similar. Going from English to German is much faster. But if you already have a second language, it can help learn a third. If you have English and Spanish, French might be faster than if you had just one of them. On the other hand this still depends on the distance between what languages you have and what you're learning. I find it unlikely that knowing English and Arabic will help you learn Japanese much faster. There might be some kind of pliability thing going on though, where having the experience of learning a language in general can make the process easier a second time.
Also if teaching a baby a language early is measurably better than later, can we teach a baby sign language? If the language learning bottleneck is voice, can we use hands?
It's already been mentioned in several comments, but the common thread that seems to be getting overlooked here is that children, as a group, are operating in a reliably different learning context than their adult counterparts. I'm going to summarize the effect as being because, as a kid, it's your job to learn and, the younger you are, the more true this is.
It's just difficult to boil that down to a small set of metrics to measure the effect.
Related: https://www.ted.com/talks/sir_ken_robinson_do_schools_kill_creativity?language=en
I speak excellent Spanish, mostly acquired as an adult by immersion in a couple of almost English free environments. My pronunciation and syntax will never be mistaken for a native speaker and although I understand grammatical gender perfectly, I cannot consistently make the correct reference. My daughters who grew up in a mixed Spanish-English home have native-language facility in both acquired in exactly the same amount of time. "How quickly" is not exactly the right question.
My girlfriend has been an English tutor for Russians. She's worked with both adults and kids. This experience made her confident that "children learn language faster" is a total myth. Children are stupider, less able to concentrate and pay attention, often forget things that they have just studied. Whatever advantage they have due to their brain placticity, it's completely compensated by everything else.
I think the critical period for phonemic perception is probably more real than this indicates - possibly not that it's absolute, but at least that "learning" strange phonemes is incomparably harder . My experience with linguist geek friends in the con-lang world is that you can be coached to articulate whatever phoneme, but coaching efficacy plateaus when you can't hear the difference between the target phoneme and other nearby phonemes. Experiments that focus on accent (phoneme production) are missing this, and that's reasonable because the overwhelming majority of phonemes in dominant languages (which people are learning in experiment-tractable quantities) are pretty straightforwardly aurally perceptible, but con-langs get into weirder phoneme qualities.
Anyone that tried learning their 2nd or 3rd or Nth language as an adult knows that there is a big hurdle in acquiring a new language that is accepting that you will speak and think like a child for a long time. We feel frustrated, dumb and are tempted to revert to our native language to recover our self-image of a capable adult, thus squandering our acquisition efforts.
I often feel extremely frustrated communicating in English (my 3rd language). Even writing this post is a little painful; I'm much more eloquent in my head, and in my native language, than this!
Maybe because children don't have the same crystalized self-image as adults they will just soldier through the middle stages of a second language acquisition which is when adults often become frustated and lose motivation.
Had to skim the article so I could get to work, but I am a speech language pathologist. As well as making some of the points Scott makes (Adults who have no language exposure don't develop language fully, the deadline for a native speaker accent is puberty) my professors also argued that early language intervention for kiddos with delays and disorders actually saves money by reducing later services that child will need.
I became friendly acquaintances with a woman from Russia who had moved to America with her husband and child. They liked it here, especially the shopping (Costco! She couldn't believe I had never been and insisted on taking me). Not exactly consumerism in the American "house proud" way, though. They lived a bit like college students, and indeed she retained an idealistic peacenik quality, the one thread remaining of "Communist youth". They were very bright - he though not having even owned a car in his life, quickly got into racing, I'm not sure what you call that sport - where you work on the car in your driveway, souping it up, then put it on a trailer and drive to races. He had a successful global business. I had the impression that though they had left with official imprimatur, even for the cat, after awhile it was not strictly advisable for them to return and I think they did so but once.
He never *seemed* to become very comfortable in English, though his taciturnity in my presence made it hard to tell. He may have found Americans (me) banal, and thus not bothered to reveal his comprehension outside of business.
The daughter thrived in school and was almost unimaginably quickly fluent in both language and kid culture.
My friend was interesting, though. Rarely pausing, when she spoke, though that may have been partly frustration borne of the language barrier - our conversations were generally sort of one way. I don't think she ever asked what my husband did for a living, for example, though it actually would have interested her, given what I soon learned were her sympathies. Maybe it was my fault. I *am* a good listener, eager to scavenge other people's lives, and it can be like undamming a stream. She was overflowing with things to say, me trying to unlock the meaning. (The other two would have moved on with an air of, this is boring, Mom, can we go?) What I found fascinating was that she routinely went to the library, from the very first, and checked out books that I would categorize as "difficult" in terms of reading level. Despite being scarcely able to utter a sentence and having learned no English in school. There was a used book store too which she pillaged in a way that suggested books had not been so surplus in Russia. She took home armloads, and read them. Once she came and fed my cat while we were away on vacation, and when I returned I found I had a new copy of "Master and Margarita" because for some reason she knew she didn't like the translation on my shelf.
(I didn't much like M&M, found it a bit of a mess, especially the last third, but keep it as a memento.)
But throughout our years of acquaintance, our conversations never fully evolved from pantomime and guesswork and misapprehension.
Once, for instance, I had casually asked her the name of the town she was from.
I became almost exasperated when despite a torrent of words five minutes later I still didn't have the name of her town.
Five minutes after that, she was chalk-rock drawing a map of the Soviet Union and I was beginning to understand the nature of her parent's work. "So *not* medicine, and the town had no name ...?"
I moved away and miss our " chats". (Though before that he'd rather improbably taught her to drive and bought her a German luxury sedan, and, American style, she ceased walking and biking everywhere.) She mentioned very early on - indicating the houses all shut up, TV screens flashing, that I was the only person in the neighborhood who did what was customary back home in the evening: stroll about the streets at leisure, ready to talk at length with people.
I guess I miss Russia.
Hi Scott - did anyone contact you regarding the "Language Learning" idea posted on Reddit a few weeks ago? I've been trying it out on Don Quixote, but I'm 1) the world's least qualified person to implement it and 2) very slow at doing anything.
I'd still be interested in hearing your more complete thoughts.
I suppose I'm now in the minority, at least among admirers of Scott, in thinking that "Critical Periods For Languages" include those distinguishing abbreviations such as "et al." and "e.g." from entire words. Omission of the periods is a mental speed bump for me as a reader. On QWERTY keyboards, the period is easy to type.
For whatever it’s worth I’m a 32 year old who has been learning Chinese for almost exactly two years. I study about 20-30 hours a week. It’s remarkable how much easier it is to learn to read than it is to hear and speak.
I have no way of knowing for sure but I guess I’d estimate my CEFR level is around B2. I don’t have any particular reason to doubt I’ll be functionally fluent in the next two or three years at this pace. The Chinese people I speak to on italki and Preply tell me my progress has been pretty fast and I sound fine but honestly I have no way of knowing whether they’re just being kind.
I wish you would’ve spent some time exploring the effect of IQ on language learning. There’s some interesting research on this topic but I think it would be a fun topic to read your thoughts on.
Not sure if someone addressed this, but learning a language after 30 is harder because your *entire life* is more “solidified”, for lack of a better term, in many ways that make it hard to drop everything and immerse yourself. I studied Russian and made little progress until my abroad semester, where my ability improved dramatically. But I was 20, and the important thing here may not have been my brain but being in a place in life where I was both able and expected to drop everything and devote myself fully to language learning. My programs also had tutors paid to talk to me.
After 25 most people don’t have those kinds of opportunities. At 30 I moved to the Netherlands, and noted that the only Americans I observed becoming genuinely fluent in Dutch had lived with Dutch partners, usually in their 20s. At 30 I had already met my (also American) husband. I had a job and faced the general difficulty 30-somethings feel in meeting new friends in new cities. The Dutch 30-somethings typically *had* enough friends and family to make them too busy to spend too much time with me, definitely not the kind of time where they might want to struggle through coaching me in halting Dutch. I paid for lessons, but only got so far.
At 40, with all of the above compounded by children and a home, it’s a wonder anyone even tries to put in the kind of effort needed to get proficient in another language. It might have less to do with our brains and a lot more to do with the shape of our lives.
Getting a native accent is a combination of listening, to yourself and others, and phonological competence. By phonological competence I mean the ability and awareness to make your organs of speech do new things.
I got to native accent in Japanese as an English speaker by recording myself, and a phonology class I took in college gave me the phonological competence to change my accent to sound more native. There are good resources about how to physically make the sounds of different languages, it's fun to stretch and exercise youf organs of speech so you can make them.
"Older research on this topic focused on feral children like Genie, who had been abandoned or abused and so never learned language. They had a hard time learning language even after being reintegrated into society; most never succeeded. But skeptics argued these children had lots of other problems besides lack of language exposure; maybe the abuse and neglect damaged their brains."
With Genie, a complication factor is that she might already have been mentally retarded before her abuse. From Wikipedia: "When she was 14 months old, she came down with a fever and pneumonitis, and her parents took her to a pediatrician who had not previously seen her. The pediatrician said that, although her illness prevented a definitive diagnosis, there was a possibility that she was mentally retarded and that the brain dysfunction kernicterus might be present, further amplifying her father's conclusion that she was severely retarded.[30][10][33]"
In fact, the conclusion that she was mentally retarded was partially why he started abusing and neglecting her.
Genie was also punished severely for trying to speak during her captivity, which might have conditioned her to avoid speaking even afterwards: "Researchers concluded that, if Genie vocalized or made any other noise, her father would beat her with a large plank that he kept in her room."
I think the missing element here is the incentives. Kids and teenagers are incredibly social. Some of these friendships will become life long bonds. They have a very high natural incentive to communicate well with peers, especially if they are new to the country and suddenly immersed in school and other native social life. Adults, however, only have an incentive to learn enough language to get by. Adults in general don't make as many new friends as children. They have an incentive to learn enough to be productive at work, but even that may be within a immigrant community where they can speak their native language part or most of the time. An adult all alone probably could learn as well as children if they had no choice, but in most cases the incentive is much weaker than for children.
"Meanwhile, 36-month-old Spanish children will still be barely saying their first complete sentences. Advantage: adults!"
This... doesn't match up with my experience around 3 year olds? I used to work in childcare and have known many. By that age, I can speak to most kids the same way I would speak to an adult in normal conversation (unless they've been raised on baby-talk...).
Especially if you give them a year to account for the fact their mouths are still forming and compare against 4 year olds, I'd assume higher level of fluency in the child than an adult with 900 hours practice 9/10 times.
Hi Scott!
It was a pleasure to see me linked in your newsletter this morning, even if it was just to disagree!
I'll admit I was a bit sloppy with the citations in the aforementioned essay. My context for the "adults seem to learn faster, but plateau earlier" comes from Lightbrown and Spada's How Languages are Learned (p. 165):
"In 1978, Catherine Snow and Marian Hoefnagel-Höhle published a study on a group of English speakers who were learning Dutch as a second language while living in the Netherlands. The learners included children as young as three years old as well as older children, adolescents, and adults. On tests administered when learners had been in the country for less than a year, adolescents were by far the most successful learners. They were ahead of everyone on nearly all of the tests. Furthermore, it was the adults, not the children, whose scores were second best. In other words, adolescents and adults learned faster than children in the first few months of exposure to Dutch."
And in Second Language Acquisition Myths, by Steven Brown and Jennifer Larson-Hall, the authors' #1 myth is "Children Learn Languages Quickly and Easily While Adults are Ineffective in Comparison." I take this as evidence that at least a substantial fraction of the SLA expert community believes that adult learning *speed* isn't slower than children, if we hold the environment and method of acquisition constant.
I'm not an SLA researcher myself, but from reading a few survey books on the topic, my sense is that this is a widely held view, if not universal among experts.
The problem I have had with acquiring fluency in other languages is that I think in English. I dream in English. My internal monologue is in English. Thus any other language I learn is halting to speak (though less to read) because it requires a mechanical internal translation process. I suspect those mental processes become calcified with age, even if the actual memorization and grammatical proficiency stay the same or improve. I can learn pronunciation moderately well, but actually speaking at a normal rate is the limiting factor due to my thoughts. I suppose immersion may be the only possible remedy, but I'm a little skeptical even of that. I once asked my cousin, a Frenchman who emigrated to the U.S. in college over 40 years ago (still has a pronounced accent though entirely fluent), whether he thinks and dreams in English or French. He said still mostly French, with just a smattering of English words.
I didn’t start learning a second language till I was in my thirties. By that time 10 years of exposure to mining noise had caused some moderate hearing loss. There were some Russian phonemes that I just couldn’t distinguish. I was terrible at pronunciation but clever enough with the grammar that I managed to hold my own
It’s a pretty major regret that I’ve remained more or less monolingual all my life but I've seen things you people wouldn't believe. Attack ships on fire off the shoulder of Orion. I watched C-beams glitter in the dark near the Tannhäuser Gate…
Oh wait, that was Rutger Hauer in Blade Runner. Never mind.
I do like to hail the usually Hispanic building crews working in my neighborhood with a cheery “Oye como va!” though.
If they mistakenly think I actually speak Spanish I have to pull out my only other canned phrase.
“Ese y lo que estoy diciendo ahora son el único idioma español que conozco.”
If they know enough English, we might go on talk about how much I love Carlos Santana’s cover of the Tito Puente cha cha cha classic.
"This doesn’t seem to match experiment, where the “critical period” for having a perfect accent lasts until age 10 - 12. Also, sometimes talented people who try really hard can have good pronunciation even if they start after that time."
My mother emigrated to the US from Ukraine in 1935 when she was 12 years 11 months old. They went directly to the small town in southeastern Ohio where her uncle lived and owned a couple of prosperous businesses. She went to public school in that town.
When she arrived she did not know any English. She spoke, read, and wrote Russian and could understand and speak Yiddish. She had never been exposed to any language that used the Latin Alphabet.
She was placed in first grade in her public school when she arrived, but graduated high school on time. She went to Ohio State, majored in Chemistry, received a BS, and married my father (a strictly monolingual American who had no interest in or aptitude for foreign languages). They stayed in Ohio where my two brothers and I were borne and raised. She never spoke to us in Russian nor did she attempt to teach us Russian.
In the late 1950s, Ohio State, like other R1 universities began a program in Slavic Studies (Cold War I, people). My mother enrolled in the program and earned a PhD. She thereafter taught Russian language and Russian civilization to undergraduates. Beyond her dissertation she did no academic writing, but she did do public outreach work for the Slavic department via a weekly radio show.
How good was her English? It was almost perfect General American. She could teach college courses and do radio programs in Ohio with out being misunderstood. The only trace of an accent was in the pronunciation of certain words. E.g. the oo in book came out more like boot than like look.
Her brother, who was 6 years younger than she was (6 when he emigrated), had no accent different than the Americans who grew up in the Ohio town where he went to school.
Effort and teaching have a lot to do with it. Hugh Laurie who played Dr. House on TV for 8 seasons with an impeccable American accent, was born in Oxford England, went to university at Cambridge, and played the ultimate upper class English twit Bertie Wooster in the Jeeves TV series. Daniel Craig, who played James Bond in several 007 movies. When he played a detective from the southern US in the first Knives Out movie his accent was unrecognizable. He was better in the 2nd one.
Accents involve the use of specific muscles, so those, at least, are acquired young and are hard to change.
I think in part the advantage of youth may come down to opportunity. My family moved to a Spanish speaking country for 2 years, during which time my 6 year old attended a local, all-Spanish school. By the end of the two years she sounded just like a local. At this point I had been studying Spanish for about 15 years, having started in my 20s, and she now spoke better than me.
Besides her age, the other big difference between us was that she got to spend 7 hours a day, 5 days per week, completely immersed in Spanish. The most I ever got was an hour of class a couple times a week, a few passing conversations during the week, and 3-4 continuous hours on the weekends - I worked in English the entire time.
But during my time living in Spanish speaking countries I met two adults who had started learning in their 20s who sounded just like local speakers (according to locals). One of them went to a Spanish speaking college full time for 5 years, and the other worked in a Spanish speaking office for around 8, so they both had tons of exposure. I think the older you get, the more likely it is your life and livelihood are kind of stuck in your native language, and the harder it gets to absorb many hours of a new language.
Remember Franzisco Pizzaro's first expedition to South America picked up few boys so they would pick up Spanish quickly and serve as interpreters for future second expedition, an important implication of the theory. (Note that Incas failed to use the Spanish they captured, and sacrificed them instead)
> if you try to teach adults and children the same fake language, adults learn faster whether it’s taught implicitly or explicitly.
This seems to be explained than fake languages are designed by adults and natural languages evolve, shaped by our biological instinct, largely in critical period (many grammar or morphology features that a programmer could have invented, wouldn't stick. yes, there's a problem with conlangs that features don't "stick"), so natural languages are easier for children who are in critical period.
I have to imagine there are some studies coming out of places that focus solely on adult language learners and intensive language training, such as the Foreign Service Institute (U.S. State Department) or Defense Language Institute (U.S. Military), that could inform this question some more. These organizations have a mandate to teach languages to professional proficiency in a defined amount of time. There should be some differences in results as well, as I believe the students attending DLI have already been evaluated for language learning ability while students at FSI have not - they are "forced" to learn a language regardless of their evaluated "ability to learn."
Here's a study on how personality differences affected results at FSI, for example: https://psycnet.apa.org/record/1995-09432-001
Regarding the tension in accent/pronunciation critical windows of months-old vs 12ish-yo, it seems to me like both can be true: it can be that you do start out capable of distinguishing all phonemes, lose "unused" ones within months, but are able to "recover" most of them until a later age.
My own anecdotal experience supports this in that I was unable to distinguish the sounds of b/v until I learnt how to write, but was then able to painstakingly learn this distinction (probably around age 5-6)
Regarding the tension in accent/pronunciation critical windows of months-old vs 12ish-yo, it seems to me like both can be true: it can be that you do start out capable of distinguishing all phonemes, lose "unused" ones within months, but are able to "recover" most of them until a later age.
My own anecdotal experience supports this in that I was unable to distinguish the sounds of b/v until I learnt how to write, but was then able to painstakingly learn this distinction (probably around age 5-6)Q
This is really interesting, bearing in mind that language acquisition isn't the only "critical window" theory out there. Basically any stage theory of the development of a psychological trait will include critical periods. An important one is the Schema theory of intellectual development by Jean Piaget. Schema theory informed Prototype Theory, one of the leading theories regarding how human beings learn and encode information into the Long Term Memory. A schema is a method of actively searching out specific types of information in one's environment, which then forms the basis of what types of categories and concepts one has. Piaget proposed that children begin learning by actively seeking out sensori-motor schema (give a baby an object and they will put it in their mouths, then throw it at the wall). When they pass through this stage they enter the next one, which uses a different set of schemas.
Language, at least to my understanding, is supposed to develop the same way. Babies learn to discriminate between phonemes before they can emit them, they babble using phonemes they hear the adults using around them before they can speak single words, single words before simple two word sentences, etc.
Each of these stages functions much like a critical window, in that an individual must master the earlier stage to some minimum level before they have the skills to even begin the next stage. All this is supposed to be based on brain development, which doesn't occur evenly and incrementally, but in spurts, by different regions.
But whether young children have some sort of advantage over adults in acquiring these sets of skills isn't actually predicted by Piaget's theory, at least to my recollection. Perhaps brain plasticity declines with age?
Children also have an enormous comparative advantage in learning language i think. This is not the question that everyone asks but is probably more important in making life choices.
If you take school as a given, i think its clear doing immersion languages is way more productive than most other things a school might realistically be capable of.
And as an adult there are many useful skills other than language you could spend three years honing that you have the prereqs for but children dont.
Why do so many native English speaker believe that learning languages is important? They don't have to learn any language. The world already speaks theirs.
I'm asking this as someone who speaks English as a second language. Had it been my first, I never would have taken the trouble to learn another.
A very long while ago, when I was studying psycholinguistics and language acquisition, I remember reading that a very young child acquiring their first language is actually learning about the world -- they are learning to categorize the things they see and experience, put them into similar groups, distinguish among non-similar groups and make connections between things. The way they do this is through language: they name things and concepts and the links between things and concepts. This is the related to why children seem to learn language slower than adults. Children are not just learning language; they're learning about the world at the same time. Adults have already learned about the world so learning language is just putting new labels[1] onto things they're already learned about when they acquired their first language.
[1] I don't mean to imply that learning a foreign language is just a matter of memorizing vocabulary. Of course, it's much more complicated that that.
What about the phenomenon where adult learners tend to forget languages as soon as they stop using them (though they come back with use), while they remain constantly available to child learners?
There does seem to be strong evidence for a "critical period" for first language acquisition in early childhood. Cases like Chelsea and Genie suggest a lack of early language exposure leads to permanent deficiencies in grammar and fluency. For second languages, the picture is more complex. Adults and older children can learn quickly in the short term due to their more developed explicit learning abilities. But long-term, older learners plateau at lower proficiency levels.
Children can develop perfect pitch while adults can’t. The best adults can do is learn relative pitch really well. Can’t help but think there is something similar going on with language acquisition/accent as well.
I wonder if there are two effects going on. The first is biological, your brain gets old.
The second is social/exposure. In high school there is a lot of pressure to fit in, and probably not many if any immigrants who speak your native language so you probably end up listening to American music, watching American tv, and spending time talking to American friends.
College is a less communal experience that is typically large enough to have enough foreign students you can hang out with. So you get far less exposure to American culture and language.
One very important factor is how eager your interlocutors are to teach you their language. Obviously parents are extremely motivated here -- you're their child, and they want to communicate with you! But as an adult learner, I found it easier (and more fun) to learn Hungarian, where they were delighted to see me struggling with their language, than German, where they just switched to English at the first sign of difficulty.
Tangentially, it appears that for learning "perfect pitch" (as commonly known), there IS a critical period, and more tantalizingly, that period can be re-opened with certain drugs!
https://www.frontiersin.org/articles/10.3389/fnsys.2013.00102/full
Warning, anecdata incoming: I am a fully proficient (C2) English speaker. I have learned English in primary and high school and I am using it at work daily. My son spent three years (5yo to 8yo) in the USA and he is fully bilingual. I definitely see a qualitative difference between us. For me, speaking English is more effort even when I am doing it fluently. For him there is no difference. For example, I have to explicitly listen to understand lyrics in a song in English, while it just happens in the background for a song in my primary language. For my son, both languages are processed in the background.
This was a great post, but I looked at this guy’s list and I annoyed by #4. “The world around us is explained entirely by physics”. This might be trivially true in the sense that fundamental physics describes the basic stuff of which the world is made up of, but by that logic so does Democritus’s theory of atoms. Tell me how the standard model explains the specific behavior biological organisms (let alone human society) in a way that doesn’t render the word “explain” pretty much meaningless?
Adults are extremely unlikely to get a native accent, and they may plateau around 99.9% on grammar. But they can otherwise get good enough to do university-level work or hold down a white collar job. Adults can also massively outperform children who are raised from birth in bilingual households, but who are otherwise living in a monolingual community.
Most people, child or adult, learn exactly as much as they need to function.
Source: I learned French in my 30s, and I have a certificate from the French government which says I'm tolerably good at it—good enough to go to a French speaking university, at least. I personally know other adults who worked even harder and who are officially qualified to go to law school. After that level or proficiency, the French government stops testing.
One thing that children have a big advantage on is a fully native accent. That window starts closing around 6, and is gone for virtually everyone by 10 or 12. The other thing that children have an advantage on is the last, say, 0.1% of grammatical accuracy. Even extremely good non-native French speakers make a higher level of gender errors than natives. (There's a super neat paper on this but it's in French.)
But if you'd be happy merely being able to take university classes or work a white collar job, then that's a totally achieveable goal. For fun, I've done long pair-programming sessions in French and it's totally doable.
Schooling from age 12-21 is one critical factor. You're required to read, to interact with your peers, and to write. All of those term papers and dorm-room bull sessions make a huge difference. And of course, that many years of schooling will build you a peer group of native speakers.
Where most adults fail is that they build themselves a "native-language bubble" and only leave it when forced to do so. Learning a language to a high level is hard work. Never underestimate the ability of a resourceful adult to avoid doing that work.
My main challenge in my 40s is keeping my academic French alive. I simply don't use it enough. My "household" French is entirely automatic and doesn't really feel much different than my English at this point.
Interestingly, the "bubble" technique works in reverse. You can build a bubble of your target language. And most rapid learning techniques do some variation of this. Once all other possibilities are exhausted, adult brains will finally give in and learn the damn language.
The key question is one of goals: Do you want to be statistically indistinguishable from an educated native adult? Or would you be happy being able to do professional work, socialize easily, and maybe even give an entertaining presentation? All of the latter are achievable, at a price.
Have I invented this, or did I read the headline findings of a study which said that kids who grow up bilingual turn out practically immune to Alzheimer's? Or it may have been Parkinson's.
I knew a family with two kids who were quad lingual. English, Urdu, Farsi and French.
This was a good post, but it should be pointed out that you clipped the quote in a somewhat misleading manner. The full belief in the original post is "Children don’t learn languages faster than adults, but they do reach higher levels of mastery." This neatly sums up the issue and the last part is just as relevant when making educational choices. Far more dubious is point 5, The reason that calorie-based dieting doesn't work isn't because 'your brain has specific neural circuitry designed to avoid starvation and, by extension, any rapid weight-loss'. It's relatively easy for fat people to lose weight fast. What they struggle to do is not to put this weight back on straight away after they stop starving themselves and this is fundamentally a hormone imbalance issue.
I feel like this seriously misinterprets the main claims about the "additional magic language ability" but don't feel the energy to write it up so just register it.
I am related to a man who served in the US army as military intelligence. He got recruited out of college specifically due to his proficiency picking up languages.
He tells me that over his career, he learned 11 different languages, all of which he used to great effect in all his different postings. He also tells me, that he forgot all of them very quickly after he stopped using them.
Meanwhile, I haven’t spoken French except sporadically in 15 years but slip back into it with some very brief rustiness whenever I must speak it. It’s a language I learned in childhood.
Anectodal, but useful