Hacker News Reader

Comment by ebiester | Hacker News Reader

I am confused at this: the window that people speak about is largely one of having any language. Aissam seems to have a language, albeit an idiomatic one that was developed to communicate with parents. If so, he has developed the speech pathways, even if any given language will be one of second language acquisition.

Now there may be another reason, but the article is either missing context or the question was not expressed in a way where the doctor answered in a way that follows the science around the critical language period, as I understand it (at least)

View on HN

Language development starts earlier than most people think. Babies babble with all the sounds the human vocal tract can make, but by 6 months they are only babbling with the sounds of the language in their household. Up to age 12 or so you can learn a second language accent-free, but at that age an important developmental milestone cements phonemes in place.

About 45 years ago I heard Chomsky speak on the idea that the human brain is wired to learn a grammar as much as a bird is wired for birdsong. So learning some grammar is innate, but the particulars are up to environment.

Source: 1/2 a century ago I was a bit of a developmental linguistics nerd. Disclaimer: But many memories have faded.

> Up to age 12 or so you can learn a second language accent-free, but at that age an important developmental milestone cements phonemes in place.

Seems... dubious? What about people who immigrate later (like in high school) and actually pick up the sounds and accents flawlessly? I've seen folks like that and I'm pretty sure they weren't in high school at age 12.

To take your point farther, I've met plenty of people who've learned their second language as an adult and can speak nearly accent free. As far as I can tell, the only difference between those with and without an accent is conscious effort.

The whole idea that there's a window that closes when you're a kid has seemed a little weird to me. Adults learn new vocab and grammar all the time. Learning another language is the same, just a little more extreme.

I wonder if there's some actual scientific evidence for the language learning window, and not just some developmental psychology observations.

Perhaps their native languages were phoneme-rich and could map those to their L2?

> people who've learned their second language as an adult and can speak nearly accent free

Perhaps it's just a capacity that some people have, similar to perfect pitch hearing. Possible, but rare.

Funny that you mention perfect-pitch because in a similar vein it seems like one of those things that can be taught but the window closes very early in childhood.

https://journals.sagepub.com/doi/abs/10.1177/030573561246394...

I was gonna bring this relation of also being a skill needing to be developed in early age, but you beat me to it.

good thing there's already a drug for that https://www.npr.org/2014/01/04/259552442/want-perfect-pitch-...

science is never fixed is it

> To take your point farther, I've met plenty of people who've learned their second language as an adult and can speak nearly accent free.

I've seen some instances of this, for instance Russians learning French.

My guess is that it's possible if the native language phonemes are a superset of the second language.

Russian doesn't have the guttural 'r' of French.

>As far as I can tell, the only difference between those with and without an accent is conscious effort.

In my completely unscientific sampling, I'm going to agree with you. This holds true too for native speakers who make no effort to improve their grammar and vocabulary. If one wants to be a better speaker, writer, and communicator, they will be by effort.

Yes, this is the real answer. The claims being made are not as articulate as they should be. You could perhaps say “before 12 a person can pick up an accent for ‘free’” but after twelve its certainly possible to pick up correct accent but it requires conscious effort, which many people are not interested in after they get their spoken L2 to the “good enough” point.

Yeah I don’t think this true at all. I spent six years in a foreign country with very different phonemes and, while I didn’t develop a very big vocabulary, I was complimented on my clear pronunciation for what I could say.

It’s not accent free, you learn the accent when you learn the language.

It's very very rare. Though it's not a cut-off at age 12, it just gets more and more rare (and takes longer) from around ten and older. Not sure if it's so much about the vocal tract, it's more about the brain's ability to hear sounds (my wife can't hear the difference between a number of sounds despite having lived in my country for many years and speaking the language well).

In all my life I've only met two 100% accent-free speakers who learned my language as adults, and a third one who was almost there. Everyone else has something I can detect. But children.. a five-year old Japanese girl could repeat everything I said with perfect pronunciation and intonation, first try. Slightly exceptional girl perhaps, she learned the language in a very short time.

> Though it's not a cut-off at age 12, it just gets more and more rare (and takes longer) from around ten and older.

I mean, nobody is disputing "it gets harder as you become older". I totally believe that. Lots of things gradually become harder as you grow older, and language doesn't seem particularly different in that regard. The question is whether that's because your body "cements phonemes in place" around age 12, or whether there's something else at play that's likely gradual and not such a sharp boundary. The fact that it's rare might be just due to the (a) effort required to learn something new being higher in general, or the (b) perceived RoI being lower, or a ton of other factors that don't boil down to "your phonemes are cemented in place"... right? Anecdotally I know in at least one particular case that I observed and inquired about, that person (who's also very smart and hard-working in general) told me they made a very deliberate effort over a handful of years to improve their accent after immigrating, and that's how they sounded like a native now. I totally believe that many people are just unwilling to invest the effort required (which certainly increases with age). I'm just finding it hard to believe there's some biological force preventing you from doing it past age 12, given I've seen otherwise.

> it's more about the brain's ability to hear sounds (my wife can't hear the difference between a number of sounds despite having lived in my country for many years and speaking the language well).

That might be true for some sounds for some people, but I also have a hard time believing it's such a general thing to the extent you're painting it here. It seems more likely to me the explanation is something else, like maybe nobody has managed to give her a good enough explanation as to how they're different sounds. (Maybe not the best example, but I had a hard time distinguishing ch and s in German until someone explained to me how they're each pronounced. Now I can hear them much better, and pronounce them not-too-awfully too.)

The sounds my wife can't separate are sounds which aren't separate in her native language - the mapping is trivial. It's the same with others from her country, it just varies between individuals (but strongly correlated with age, and to what exposure they've had to other languages when younger). And believe me, it's not about putting in the effort. But to distinguish certain sounds my wife has to watch my lips - this is particularly noticeable if I dictate and she writes. As for training - she received two years of training (30 hours a week) with expert teachers who knew a lot of tricks for how to hear and (not the least) pronounce sounds. Tricks that I didn't know about.

But I've also seen this with American and some English adults trying to learn Norwegian - a great many of them can't hear the difference between vowels which, to me, are totally different. Can't hear the difference between the words "har" and "her", for example (NB: Norwegian sounds. Not English vowels). It seems to take a couple of years of daily ear training (or rather, brain training). As always, there are exceptions. But those exceptions are truly standing out.

(Added: As soon as there's context or visibility the problem is much reduced - but it's still there, as soon as there's only audio and their language level isn't good enough to "select" the right words from context).

I've noticed this within accents of the same language. My own English accent has a very mild separation between the i and e vowels in, e.g., bitter and better. To my ear they sound wildly different, but many native speakers from other places and more English-as-a-second-language speakers can really struggle to tell apart what I'm saying, to the point that it's usually easier for me to switch languages to the local one to distinguish the words if the context is ambiguous. Oddly, despite noticing shifts in my accent having lived abroad for some time now, this is something that hasn't budged.

Similarly in my (learnt as an adult) second language, there are a couple of vowel sounds that aren't in English and I usually have to really focus to hear them, and to pronounce them correctly.

My favorite example in English is this sentence: "I peered at a pair of pears on the pier."

As a native speaker, the sounds in peer/pair/pear/pier are slightly but detectable different. But non-native speakers can almost never say or hear these differences.

As a native English speaker (American) I think "peer" and "pier" are pronounced the same, and that "pair" and "pear" are pronounced the same, but that the 2 groups are pronounced quite differently.

The classic American example for me was always "ten/tin", "pen/pin", which in my Southern neck of the woods are typically homophones. My girlfriend is Midwestern, and definitely pronounces the "e" more in those variants.

https://www.youtube.com/watch?v=qXxV2C1ri2k

The ability to distinguish sounds in an L2 that is not in the L1 is absolutely possible. It comes with a combination of massive input and being taught the tongue positions to allow the person to begin producing the two sounds. If someone says they can’t distinguish the sounds mostly it’s because they haven’t had enough comprehensible audio input.

My anecdotal example: Italian living in Catalonia (for a few years now, but moved as an adult), I speak both Catalan and Spanish with no Italian accent and what's even "worse", I speak Italian with Spanish accent ^^;

I know these languages are all very similar, so that helped a lot for sure, but the 12 years rule is definitely not absolute.

Anecdotally, I know a Serbian who went to study to the US for a year in high school and has a perfect Californian accent.

> perfect Californian accent

Oh man, I hope they sound like this:

https://www.youtube.com/watch?v=dCer2e0t8r8

Nah, he totally, like, sounds like a Valley girl

There you go!

It's bullshit. I can speak Dutch without an English accent and I learned the language in my 20s, having 0 exposure to it in my youth.

There are to main sounds which are in Dutch but not English ([ui] and the hard [G]) and one which isn't in some English dialects (the rolling [R]). However you can absolutely learn them as an adult. It just requires serious training (years of hard practice, same as with sports). You literally have to build up the facial muscles.

One thing I will admit: I choose not to pronounce the [ui] sound properly due to a combination of being lazy, identifying as an Dutch-as-second-language speaker and because for some absolutely irrational it sounds really childish to my years. That latter point played a surprising role in my lack of ability with the French language. It feels theatrical in the way that certain queer people choose to project their speak and I do not want to project as being something I am not (or be confused as someone making crude n-phobic caricatures which would be 1000x worse because it could make someone else feel insulted). Honestly it's probably a tick I have from being raised to be a polite British gentleman :)

I've noticed that some people speak foreign languages with what I consider to be quite poor accents, but they enunciate strongly. If they are smiling, it almost comes off to me like they are making fun of the language or just pretending they can speak it well. To the point I've laughed along with them before realizing my mistake and cringing.

And then I wonder which of us is more fluent? The one who with the better accent or the one who can more confidently project their thoughts in that language regardless of "skill"? (Probably the latter)

Yea but you are talking about two languages that as you correctly pointed out only a few phonemes are different, with very similar phonology, stress, etc. They are very closely related.

I feel like we're so deep in the thread - where after people have time and time again disproven the original comments message about how being 12 years old is some magic cut-off point to language learning - there are now people who just keep adding different requirements, leading this conversation nowhere and making the new point unfalsifiable and - by extension - unscientific; whereas the original comment had an allure of having a modicum of scientific rigor behind it (which, in the end, was disproven)

I think that the original comment has some anecdotal truth - because the vast majority of people do not spend the effort required to speak their second/third languages without an accent. You have to do silly exercises and stuff for very little gain in the end. You're better off learning a 3rd language.

Professional Actors do it all the time, they spend a lot of time practicing (and I assume money) being trained by Speech Therapists to different perfectly.

Indeed. Sounds like another 'replication crisis' paper in action.

To be fair, they did qualify the statement with "or so".

This isn't 12 vs. 12.5 I'm disputing, it's like 12 vs. >= 16.

> Up to age 12 or so you can learn a second language accent-free, but at that age an important developmental milestone cements phonemes in place.

This is only true if the second language has sounds that you don't have in your first language.

Oh, right. I forgot all those Dutch speed skaters who used to train in Norway a couple of generations back. Ard Schenk.. Kees Verkerk and many others. The majority of them learned Norwegian completely accent-free (you really had to listen to figure out they weren't native), despite not learning the language as kids. I figured that must have been because Dutch has such a variety of native sounds that it basically covers Norwegian language sounds.

Can you name a language pair where that works? I can’t think of one.

https://linguistics.stackexchange.com/a/30633

If your native language is Ukrainian, when you learn Russian you don't get more phonemes, unless you count slightly different mouth positions of vowels.

My personal experience actually maps to this exactly and, when I’ve encountered it, it’s been a far more predictable rule than I would have thought. This is anecdotal I know, and I’ve even looked up the science on this and it seems like isn’t as black and white as my experience would indicate.

But by now I’ve actually asked this of prob 20-30 people. All of them who came to the US before 12 (or attended American international schools overseas) had no accent. And all of them except two who had come here at 14 or later had at least a hint of an accent. There are the few like the one Hungarian I met who had no English other than spending 3 months in the US and whose English was so spot on I actually thought he was American.

In my experience it does seem that there’s something about the brain’s plasticity that changes around 13ish. For example, I started programming young and also had took physics early at my local college and seem to internalize those much better than, for example, the follow-on physics course I took later on.

But if anyone knows the science better feel free to correct me! A neuroscientist I am not…

This is well known and well studied. I highly recommend Sapolksy’s lecture on this: https://youtu.be/SIOQgY1tqrU?feature=shared

Start at 12:14 for the relevant topic, but the entire lecture is a good watch.

Relevant This American Life episode:

> Jiayang Fan has this theory that because she's spent so much time thinking about her own accent when she speaks English, she believes that when she hears other Chinese-Americans speak, she can tell how old they were when they immigrated to the U.S. (7 minutes)

> We test Jiayang Fan’s self-proclaimed special skill by having her listen to three Chinese-Americans speak, and then guess when they came to the U.S. (20 minutes)

https://www.thisamericanlife.org/786/its-a-game-show

Oh, rad, I didn't know accents worked like that! So if I speak to my kids in English, and my wife speaks to them in Finnish, they'll get to grow up with a disconcerting mix of newscaster English and northern Finnish drawl.

I wonder if getting exposed to a bunch of languages as a kid is why I have a (relatively) mild accent in Finnish now, despite only starting to learn at 26 or so.

For the two-language trick to work, both languages need to be equally useful for the kids. My sister and her Canadian husband tried having her speak to them in Norwegian, and him in English, but as soon as they understood that she could speak and understand English, but he could not speak or understand Norwegian, the kids switched to English. They still understand Norwegian, but they speak little of it and with a heavy English accent.

https://www.scientificamerican.com/article/evidence-rebuts-c...

https://blogs.scientificamerican.com/cross-check/is-chomskys...

that's not much of a defense

"Another problem with the claim that Chomsky’s theory of language “is being overturned” (as if it had ever been accepted, which is not true), is that it’s not clear what “Chomsky’s theory of language” refers to. He has proposed a succession of technical theories in syntax, and at the same time has made decades of informal remarks about language being innate, which have changed over the decades, and have never been precise enough to confirm or disconfirm."

There are a bunch of ideas that are more core and strongly supported (language is innate) which you use to explore more tenuous ideas about what the implications are and how they specifically manifest. Linguistics is an extremely nascent field compared to other sciences, Chomsky calls our stage of understanding "pre-Galilean", no one has claimed to have solved the basic questions yet so it isn't surprising that anything other than the core ideas are in constant flux. I haven't seen a good counter argument to the core ideas of universal grammar (or the minimalist program) and to refute an idea you need to actually present a counter-argument not simply say some sub-hypothesis has been refuted in the past so every fundamental idea has been refuted.

The SciAM article you linked doesn't understand the arguments Chomsky makes when "refuting" them (e.g. they erroneously say that superficial differences between languages show that there is no universal grammar).

I thought I read an article one that some very untouched tribes (ala amazon?) have fundamental different ways of communicating that undercut Chomsky's notions of universal grammer. Which the SciAm article glances on, but doesn't really go into any depth.

That's Piraha and Daniel Everett's work. Even assuming the results are correct (that Piraha doesn't use recursion) that doesn't demonstrate anything relevant. What you want to show is that Piraha people don't have the ability for recursion (i.e. they are an example of people without this innate language faculty), not that they don't apply it. I don't think anyone believes that if you take a Piraha newborn to New York they wouldn't learn English.

Chomsky explains it himself: https://www.youtube.com/watch?v=u6Lk79bnUbM&t=1386s or here https://youtu.be/c6MU5zQwtT4?si=A9t8d0oXV4dOLZTe&t=3008

It's because you lose the ability to form new abstractions once your neocortex shuts down. The sensory areas begin shutting down quite early (around five) for most. You can only hear which category the sound belongs to + the error ("accent" in this case) and lack the ability to perceive finer nuances after it happens.

Yeah, this seems confusing. Obviously his brain has a solid understanding of language, and he even recently learned a new language (Spanish Sign Language) so learning another language should be possible.

It sounds like the researchers are saying there's something special about learning spoken language. But it seems to me that there can't have been many cases similar to his.

> It sounds like the researchers are saying there's something special about learning spoken language. But it seems to me that there can't have been many cases similar to his.

We've learned a lot from people who have received cochlear implants at different ages. Earlier implantation is strongly associated with functional spoken language use and fluent speech. There's a big benefit before age 5; a large proportion of those implanted before 24 months basically have normal language skills, while few after age 5 ever fully "catch up."

edit: Here's a study of prelingually deafened adult outcomes with CIs https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5720870/ All of those studied had acquired spoken language before implantation and had some degree of effective hearing earlier in life, so were not fully deafened before the language acquisition window.

The implants provide an improvement of quality of life but do not allow most of even this population to e.g. understand spoken language on TV without subtitles.

Listening to TV conversations is a high bar. In conversations people can ask for clarification or to slow down etc.

Which explains: Before implantation, 7% of the patients were able to have telephone conversations. vs After implantation, 60% of patients are able to have telephone conversations.

Also, the technology dramatically improved over time so we don’t have long term data on high quality implants.

Still, pay attention to context. This was people who were successfully using hearing aids and oral language before implantation. Even in this subpopulation, they did not do nearly as well as children do, even though this subpopulation was less deafened than most deaf children.

> Also, the technology dramatically improved over time so we don’t have long term data on high quality implants.

We do have enough series to know that 5 year olds receiving treatment have (on average) significantly worse outcomes than 18-24mos.

For the timescales we have tested there’s significant differences, and in terms of quality of life it’s clear early intervention is a significant benefit.

However, slower adaptation isn’t zero adaptation. The limits for people implanted at 5 when they are 50 is still an open and IMO interesting question.

> The limits for people implanted at 5 when they are 50 is still an open and IMO interesting question.

Slower development usually means a lower plateau, and I think we pretty much have to assume as such (and can be prepared to be pleasantly surprised).

Else, we get to wishful thinking: older people on older devices developed more slowly and plateaued at a lesser value of hearing. Now, we have implanted older people on newer devices, and they're developing more slowly, but hey, maybe they'll eventually develop fully normal hearing.

Do you have a source for that plateau? I’ve read that early success predicts future success on age adjusted tests. But the children were still improving in absolute terms over a decade post surgery.

I was arguing mostly from the standpoint of delays in development are almost always correlated with lower ultimate attainment, no matter what measure you're looking at.

But if you want cochlear implant specific data, here-- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10760633/

You're right that time narrows the gap between early implantation and later implantation, but the slope of that narrowing is pretty small by the 20 year mark (and barely statistically significant in this moderately-sized study) and the gap is relatively big.

The difference of time of implantation between the two groups was relatively small (mean implantation at 45 months vs. 34 months) and produces a gap that's durable for decades. >130 months is way, way, out from 45 months.

Thank you, interesting read.

Looks like scores for both groups are still improving at 25 vs 20 years so gap isn’t closing. I was expecting people to max out what the hardware is capable of or reach normal levels, but that doesn’t seem to be what’s happening.

I was surprised at the amount of continuing improvement, too.

Very interesting paper!

Still, lots of work to do, to quote "However, it was not possible to control other factors, such as the socio-economic environment of the participants.".

In my view this could affect the study quite a lot (or not, but unknown for now). They mention the initial intervention was 3 months on-site, but after "the patient returned to his area of residence, where he/she would have speech therapy and special education".

Never fully catching up sounds a lot different than "window for learning spoken language is permanently shut". Am I missing something?

"Permanently shut" for everyone is probably an exaggeration.

A better description "very, very few of those [with hearing restored] after age 5, who had never had any hearing before, develop anything close to normal spoken language skills."

I recall that they trained a part of an insect's head like the nasal passage and it was able to be used for language better than the model at that time chatGPT2. So there's something innate in nature that can learn human languages.

Can you find that study?

Took me a few hours but it was fascinating. I couldn't find it again easily.

https://www.discovermagazine.com/the-sciences/fruit-fly-brai...

https://oadoi.org/10.1371/journal.pgen.1007430

Cool, and thanks for search!

I think you may have slightly over-sold the study, though -- or at least what you remembered from it.

My reading of the first study was that they took a simulated version of a relatively large (2000 node) neutral network that makes up part of a fruitfly brain, and were able to do standard neutral network training on it to do some language prediction.

I'm not sure that this says anything about fruitfly noses being wired for language though. I expect that they could have taken that same simulated architecture and trained it to do anything that regular neutral networks could do -- detect faces, make stock market predictions, play a (poor) game of Go, or learn a homeowner's thermostat patterns.

I think it's just more a statement about the power of neutral networks in general.

I did misremember, it seems they used a network of algorithms based on how flies use 2000 neurons and the original article I read it from a while ago may have oversold it. The fact it was able to predict well is them looking for inspiration in insects, and that they'll be using more insect inspired behavior.

There was very sad case of an abused girl named "Genie" who was never spoken to by her parents - kept in a locked room as a toddler and she was rescued at age 13 and made a ward of the state but never learned the ability to speak despite best efforts of linguists and therapy.

It is if the basics of language and fine motor control required of the vocal cords/tongue/mouth in her brain was not primed and that ability was forever gone.

Sensory and motor systems have critical or sensitive periods early in life and then the neural network is pruned and the critical period closes. This is why second language acquisition is much easier at younger ages. There are some drugs that may reopen the critical period although not much can be done about the pruning. So basically all the early neural circuits for interpreting sounds are not organized correctly and it's unclear if they can be post critical period.

Maybe the brain's plasticity. I read somewhere, a long time ago, that brain nerve cells wonder around and do their thing based on incoming stimuli when the child develops. During very early age a lot of things can be corrected by the brain if something goes sideways, thanks to the plasticity of these nerve cells, but after a while, when the nerve cells "settle" and "conquer their role", they cannot change anymore.

Audio processing is a bit light sight processing - once you miss the critical development period, which is not directly related to the "critical period" for language learning, you will never actually develop them. People can develop something, for example people who had oxygen-destroyed corneas causing blindness who later got corneal transplants, but it will never be vision as you know it.

Search "kitten vertical lines experiment" on the Famous Search Engine. Kittens not allowed to see horizontal lines for the first few months would never be able to see horizontal lines, ever.

It's at least somewhat like that with humans too. Ever wondered why some kids are wearing a patch over one eye? If the child needs glasses but (in particular) when they didn't get proper correction early on then they may have double vision, and what the brain does is to block one eye. That eye, despite "seeing", will lose the paths in the brain necessary for seeing well. The patch forces the brain to start using the eye again.

This happened to me - the doctor told my father "no need to check this regularly", and after some years one of my eyes had indeed lost resolution. It's still like that. One eye can see very well, the other at much lower resolution. Though I found that even at middle age it was possible to improve that to some extent - not the actual resolution, but the brain's ability to actively use the eye could be improved a little. I would read books with only one eye. Could only read half a page at the beginning. But it's impossible to recover the vision I lost as a child, which was caused by the brain ignoring the eye.

I have the same however my other eye is fine with regard to resolution - I just need glasses or a contact lens.

But yes, the binocular vision is permanently shot (though I get some improvement at times).

That's what you get for being an insufferably stubborn kid.

I have the same situation.

The annoying part is that my "untrained" eye is not near-sighted, but my "trained" eye is. I suspect it was different in the childhood (untrained: far-sighted, trained: not far-sighted) and then shifted in the direction of near-sightedness over time.

The ad-hoc language (generically called "kitchensign" apparently?) might be too primitive.

Reply on news.ycombinator.com