[alecst]

Interesting theory. But it cannot explain a few things. One of which is why children never have accents, and adult non-native speakers do. Another is that children do not need lessons to learn a language, and adults always do.

[moron4hire]

I'm not quite sure what you're trying to say by "children never have accents", but children do have accents, both the accent of whatever locality in which they live, as well as their own accents from not learning the language precisely. My 3-yr-ld son pronounces all words end in an "-ar-" sound of some kind as "-aiee-". He wanted to play "cards" and it came out "cai-eeds". There are other, similar foibles in the language of children that I think could easily be called "children have a unique accent".

Also, it is very not true that children do not need lessons to learn language. If anything, children receive MASSIVE amounts of explicit language training that we would never think to apply to adults. Children have songs about the alphabet and numbers. We play games with them about colors and shapes. Before the age of about 5, almost all of their toys are fundamentally designed around learning components of language. All of the books that we read them are about.

Both my 3-yr old and my 5-yr-old make what I find to be a hilariously cute error in speaking. Things that belong to them, they say are "Mines". I thought about it, and their way is more consistent. You say that toy is yours, hers, his, theirs, ours. It's only in the 1st person that we drop the -s sound at the end.

When do children gain fluency? How do we even define fluency? In the language training industry, we have the International Language Roundtable Scale (https://en.wikipedia.org/wiki/ILR_scale). The ILR Scale ranges from 0 for a raw beginner to 5 for "educated, native speaker", with people typically appending "+" to a level to blend in between levels a little. Based on the ILR scale, my 5-yr-old is a 2 and my 3-yr-old is a 1. I know full-grown adults, born and raised in America, who would probably only rate a 3+.

Children do not learn fluency without massive effort on both their and everyone else around thems behalf. And then adults complain about having to do 5 hours of homework every week and whine about not gaining fluency in Mandarin. "It's just easier for children". Yes, in a round about way, it is easier, but those reasons are purely social. Given that some adults do demonstrate the capacity to achieve fluency, yet are not living anywhere near a completely, 100% immersed life like a child does, there is clearly some natural advantage that adults have that makes up for the lack of nurture.

[alecst]

1. Accents are evidence that the speaker did not fully acquire the phonology of the language. A French native speaker will speak English with a French accent because they haven't acquired the full English phonological inventory. Children who learn a language don't have this problem.

2. Children will learn colors and shapes just fine without explicit instruction. They do it all the time. In the pre-developed world, children didn't get taught how to speak. There were no flash cards or toys for learning numbers. They just learned by observing.

3. Letters is something else. The orthography of a language -- learning how to write it -- is a different beast from learning how to speak it. Education is necessary here. So we agree about that, for sure. No one will passively acquire how to write.

4. Children make errors in production all the time. As they learn a language they make generalizations -- generalizations that actually make sense, like "mines" -- but which are considered "wrong" by adult speakers. They'll correct themselves over time without instruction.

You might want to check out reading Steven Pinker's "The Language Instinct," it has a lot of ideas and research that might be new to you.

[codethief]

> 1. Accents are evidence that the speaker did not fully acquire the phonology of the language. A French native speaker will speak English with a French accent because they haven't acquired the full English phonological inventory. Children who learn a language don't have this problem.

True. But, referring to my earlier comment[0], this is a matter of changing language learning habits for adults, not a matter of ability.

> 2. Children will learn colors and shapes just fine without explicit instruction. They do it all the time.

Define "without explicit instruction". We explicitly read our children books that put a ton of emphasis on getting the words for colors, shapes and animals across (by repeating them over and over). We repeatedly ask our children questions such as "What color is this?", "What's that animal called?". If that is not teaching, I don't know what is.

> In the pre-developed world, children didn't get taught how to speak. There were no flash cards or toys for learning numbers. They just learned by observing.

Well, they certainly must have been listening, too, and their relatives must have pronounced the words for colors in the first place or otherwise they certainly would have never learned those words. But where do you draw the line here between "passive observing/listening" and "being taught"? To me, at the end of the day, these are all just different learning techniques and I don't see anything special in the way a child's brain acquires a language compared to an adult's.

> No one will passively acquire how to write.

Funny, I actually disagree here. I remember that, back in the day, my 4-year old cousin used to copy books verbatim, letter by letter, before he could actually read or write. (Where I'm using "write" in the sense that one puts letters on paper to form words and sentences and to articulate some meaning.) By the end of this whole process, my cousin knew how to read and write perfectly. And by that I mean: He was an absolute grammar nazi by the age of 6. Later, he would then go on to read dictionaries in foreign languages aloud, page by page, for hours while we were on vacation – just to annoy me. Today he speaks four languages absolutely fluently and he's at least somewhat proficient in another two.

Anyway, I'm sure my cousin once had to ask my aunt about the pronunciation of individual letters of the alphabet – to map them to the sounds he already knew – and probably also about words whose spelling differed a lot from their pronunciation. But if what he did is not "passively [acquiring] how to write" (and read), I don't know what is.

Then again, he put in massive amounts of time to accomplish what he might have learned much faster in elementary school.

> 4. Children make errors in production all the time. As they learn a language they make generalizations -- generalizations that actually make sense, like "mines" -- but which are considered "wrong" by adult speakers. They'll correct themselves over time without instruction.

Sure, as will adults. It's just that correcting them (whether a child or an adult) leads to much faster results and stops mistakes from becoming ingrained in their mind. Once again, it's a technique speeding up the learning process. (Which is why it's so helpful to have a native speaker correct you while you're learning a language.)

> You might want to check out reading Steven Pinker's "The Language Instinct," it has a lot of ideas and research that might be new to you.

Would you mind summarizing some of these ideas that, if I understand you correctly, undermine OP's (and – by extension – my own) argument?

[0]: https://news.ycombinator.com/item?id=27917214

[alecst]

I think I'm okay ending the back-and-forth here. I'll post the link to the book and you can check it out. It's a fun way to be exposed to modern (last 75 years) research in linguistics. Talks a lot about language acquisition research, which you have some opinions about. Come at it with an open mind and you'll learn something, I'm sure.

https://stevenpinker.com/publications/language-instinct-1994...

> Everyone has questions about language. Some are from everyday experience: Why do immigrants struggle with a new language, only to have their fluent children ridicule their grammatical errors? Why can't computers converse with us? Why is the hockey team in Toronto called the Maple Leafs, not the Maple Leaves? Some are from popular science: Have scientists really reconstructed the first language spoken on earth? Are there genes for grammar? Can chimpanzees learn sign language? And some are from our deepest ponderings about the human condition: Does our language control our thoughts? How could language have evolved? Is language deteriorating? Today laypeople can chitchat about black holes and dinosaur extinctions, but their curiosity about their own speech has been left unsatisfied—until now. In The Language Instinct, Steven Pinker, one of the world's leading scientists of language and the mind, lucidly explains everything you always wanted to know about language: how it works, how children learn it, how it changes, how the brain computes it, how it evolved. But The Language Instinct is no encyclopedia. With wit, erudition, and deft use of everyday examples of humor and wordplay, Pinker weaves our vast knowledge of language into a compelling theory: that language is a human instinct, wired into our brains by evolution like web-spinning in spiders or sonar in bats. The theory not only challenges convention wisdom about language itself (especially from the self-appointed "experts" who claim to be safeguarding the language but who understand it less well than a typical teenager). It is part of a whole new vision of the human mind: not a general-purpose computer, but a collection of instincts adapted to solving evolutionarily significant problems—the mind as a Swiss Army knife. Entertaining, insightful, provocative, The Language Instinct will change the way you talk about talking and think about thinking. New in 2007: The new “PS” edition contains an update on the science of language since the book was first published, an autobiography, an account of how the book was written, frequently asked questions, and suggestions for further reading.

[codethief]

> "It's just easier for children". Yes, in a round about way, it is easier, but those reasons are purely social. Given that some adults do demonstrate the capacity to achieve fluency, yet are not living anywhere near a completely, 100% immersed life like a child does, there is clearly some natural advantage that adults have that makes up for the lack of nurture.

I like this argument, it puts it very succinctly!

On a completely different note, seeing that you're developing VR applications for learning languages: How is that coming along? Is Diplomatic Language Services already using it in production? And what improvements in learning/teaching compared to non-VR applications (and non-VR private classes) have you seen? I mean, I generally understand the appeal of VR but I hadn't heard of it in the context of language learning yet, so I'm wondering what the advantages might be since it's a purely visual thing, so a priori not necessarily more conducive to teaching a language(?)

PS: Now that I've seen you mention your alma mater on your website: What a coincidence – I once went to Shippensburg for one of my boxing fights! :) What a beautiful place!

[moron4hire]

We're currently in trial phase, so we don't have a lot of results yet, but what we've built so far is based on existing research and the results we have seen are consistent with it. The research has shown significant results in various learning factors, primarily based in sensory engagement reinforcing neural pathways. I'm... not sure how much I can share about that research, as it was presented to us as proprietary information by a person we are hoping will be a research partner of ours in the near future. Though I will say, I'm quite proud of myself for having intuited most of the factors and built them into our design before we met them :)

The VR at the most basic level provides a context in which to learn the language. It's one thing to read a book about a place and learn the words associated with the people and culture of that place. It's another thing to stand in that place and see the thing you're talking about.

And we've all heard that full immersion is the "best" way to learn a language. But that's just not an obtainable goal for most people. As I said, most of our students are government employees. They work most of their time from the States or a US Military base. When they are put into a situation to use their language skills, it can often be at the last minute for a deployment of some kind. They're adults with careers, kids, houses, and not enough money to travel the world whenever they want. So the VR gets us closer on the "being there" scale than they'd get otherwise.

We're taking a different approach from our competitors in the market. There are a number of "VR" foreign language training apps that are basically just DuoLingo or Rosetta Stone in a 360 video (and in a few cases, we've seen people calling it VR even when there's no headset involved, sigh). You are meant to purchase a course that is solitaire and self-guided. They're all using speech recognition engines to judge pronunciation. That sounds like a great idea on the surface but turns out to be complete garbage in practice. Speech recognition engines only work about 95% of the time for fluent speakers, where the threshold for non-frustrating Human-Computer Interaction requires more like a 98% success rate. So you get into a problem of not knowing if the errors you detect are because the student isn't pronouncing the words correctly, or if the speech engine is not doing its job. Even worse, you get a lot of false positives as well, which gives students the illusion that they are progressing when they are not (I've actually done a fair amount of work with speech-oriented interfaces as well. My career has been... very broad).

In our case, we already have a successful business doing 1-on-1 language classes, so we've designed the application to fit into that. Also, I wanted to make sure that our instructor body didn't think we were trying to put them out of a job. Most of our students are in class for about 5 hours a day, two to three times a week, for about 3 months at a time. They spend that time 1-on-1 or 2-on-1 with a certified language instructor. Our VR app constitutes about 1 to 2 hours a week of that course, and it still involves your instructor. In this way, we avoid the pronunciation assessment issue that our competitors have entirely. But we are also providing a much more personalized experience for our students, as well.

We take imagery from Google StreetView, combine it with didactic content, and weave in narrative of taking a guided tour of the location with your instructor (I've even built a whole lesson editor that our curriculum developers are able to use on their own, so we're getting pretty fast at making new content, too). You discuss the place and maybe role-play a few interactions like renting a hotel room or buying food at a restaurant.

That's Phase 1. We are currently entering Phase 2, which involves different interaction metaphors and more interactive play for role-plays.

Even in traditional language instruction, there are concepts of "immersive" role-play, what are called "isoimmersions", which could be as complex as setting up a mock bank teller booth in a room or as simple as lining a bunch of chairs up and calling it a "bus". I see a lot of our product as taking that concept and making it not hokey. Cultural learning is also a big part of our training regimen, as it doesn't do a diplomat much good to learn Arabic without learning a specific region's dialect and culture.

In our own trials, we're seeing a very sizable increase in student engagement. The students are reporting and showing that they are more motivated to study. The instructors are also having fun with the project, which we think is significant. We have a pretty broad range of student level, but we're seeing improvement across the board. Even the seasoned students who are used to isoimmersions have discussed how the VR headset gives them a brand new perspective on the content.

Our team is basically me as head engineer, a junior developer, and a rotating roster of 3 language training specialists from whomever has time from fulfilling our regular work contracts.

As for Shippensburg... it's a nice place to visit. I don't miss it much. Gettysburg is a better town in that region.

[codethief]

Thank you so much for taking the time to write such a detailed response! The approach of "traveling" to a foreign place using VR is a really cool idea I hadn't even thought of! Now I kinda wish your app were open-source (or available for a fee) and my online Spanish teacher could use it in his language class with me haha. :)

[codethief]

Those things are fairly easy to explain and they also aren't god-given facts.

> children never have accents, and adult non-native speakers do

This mainly due to two things:

1. Language learners usually try to map the sounds they already know (from their own mother tongue) to the new language's sounds. In fact, there are studies that suggest that, depending on their mother tongue, language learners will not even notice certain differences between sounds in the target language and their native language. (Consider how Japanese people tend to have issues distinguishing the letters R and L, or how both English and Spanish native speakers usually cannot pronounce the French/German "r" properly because that sound doesn't exist in their native language.)

This is not at all set in stone, though, because it only takes a handful of weeks of focused practice to reset your ears (and your tongue) and tune them to a new language's sounds – preferably before you learn any of the vocabulary or grammar of that language.[0] Also, the greater the spectrum of sounds you already know (the more languages you speak), the easier it will be for you to learn a new language's accent as your brain will be already attuned to listening closely to tiny differences in sound and speech melody.

Once again, consider that a child has years to learn the sounds and all these nuances.

2. Habits. Language classes almost never focus on pronunciation and speech melody in the beginning. From my POV this is a huge mistake as it means that language learners attending such classes will sooner or later get into the habit of pronouncing words of the new language using their native language sounds. That is, when they see (or think of) a word in their new language, they will no longer pause to think about how to pronounce it – they will just do it. Unfortunately, at this point it's pretty much game over as it will take a lot of work to change these habits. Then again, a lot of people also don't really care that much about having an accent.

> Another is that children do not need lessons to learn a language, and adults always do.

This is not true and I know a few people who have gone the full-immersion-zero-lesson route. It is incredibly hard, though, given the time constraints you usually have (usually a few months to a year) and you will usually progress only very slowly. Again, just consider how much time it takes a kid to learn a language by just observing and mimicking others! For adults, lessons are simply a much faster way to get started and become somewhat proficient in a language. Also, once they've taken a few lessons, they will be orders of magnitude faster at learning the rest of the language.

[0]: For an introduction to this approach of learning a language, I can recommend Gabriel Wyner's book "Fluent Forever" and, also, his pronunciation trainers and his YouTube videos on the International Phonetic Alphabet (IPA). They got me from speaking absolutely zero Spanish to being asked whether I'm a native speaker in about two months of living in Colombia. …which doesn't mean I was speaking perfect Spanish at all – but it sounded like I was! In fact, on various other occations, people told me that they could tell precisely which city I had learned Spanish in because I had become so attuned to the accent people speak there.