emk wrote:Cainntear wrote:The ability to label concepts this way is clearly a necessary precursor to language, so we presumably have some kind of structure dealing with conceptual hierarchies that is shared with a heck of a lot of pack animals who can signal concrete noun concept with a single symbol (eg a particular food source) and even abstract ones ("Danger! danger!"). Our shared ancestor with dogs was probably 80 million years ago, but even meerkats have a system of symbolic representation of noun concepts.
Human-style language requires more than word→concept mappings.
Yes indeed, but my point is that word→concept mapping are not unique to humans... or perhaps symbol→concept is more appropriate.
I think it's pretty reasonable to assume that language would use the circuitry that allows symbolic representation of concepts. There will be structures in the brain that are about "language" that use such circuitry -- evolution might reinvent the wheel from time to time, but rarely in a clade that already has a wheel.
I believe there's good evidence that animal brains have a concept of hierarchical relations between concepts -- even to the point of generalising. Dogs can do a good job of identifying things they are supposed to play with from things they're not supposed to touch; they can categorise people on first sight into "probably trustworthy" vs "probably not trustworthy" (in a way that kind of shows where human prejudice comes from). Have you ever absent-mindedly opened a different thing in the category "kitchen appliance" from the thing you meant to? All of this probably sits quite far from the core "language centre" of the brain.
Let's take one of our closer relatives: Koko the gorilla, who famously uses "sign language".
...
Above all, Koko never uttered sentences.
Which is really my point. The existence of nouns seems to be something that existed before language.
The studies with great apes were very heavily influenced by Chomsky's concept of some kind of single "language acquisition device" -- the fact that the first "signing ape" (IIRC) was name Nim Chimpsky says a lot about the researchers' bias!
The big question was always whether language is a single genetic thing or a combination of multiple things, and I think that the ape signing researchers were so fixated on finding when the "language acquisition device" gene came into existence that whenever they saw signs of "vocabulary" they interpreted them as signs of "language", so they totally ignored the absence of anything resembling grammar.
So, sure, maybe some species have the ability to learn a few words, or "word-like" communication mechanisms. (And who knows what whales are doing.) But what is very clearly unique to humans is grammar.
This is what I'm saying, though. My suggestion is that word-like stuff is a trait that exists in an awful lot of mammals, and I believe that language uses that and is built on top of it. Mammals have tens of millions of years of vocabulary, but grammar is probably only a couple of millions of year old.
So there is very clearly something in the human brain that is determined to learn a language, and that expects a language to share certain universal features: words, some sort of hierarchical grammar, probably nouns and verbs, etc. But I don't think there's any kind of strong evidence for Chomsky's "Universal Grammar", not in the sense he means it. From my perspective, Chomsky doesn't even have the right theoretical tools to tackle this question meaningfully.
Agreed. But I think that the noun is probably the oldest thing, and effectively predates language.
I find the idea of a "critical period" for language acquisition hilarious, because my wife and I don't share any native languages, and yet somehow we've managed to talk to each other for over 20 years. Mostly we speak French, which I started after the age of 30. But my wife's climb from ~B2 to near-native English happened well after any critical period, too.
Well the critical period really is influenced by the idea of children without any language input whatsoever, which is a pretty extreme case: no L1 by 5 or 6 does seem to mean never mastering any language. I think there was a rather naive idea that this meant that adults would be worse at learning languages than kids.
I've always been of the opinion that apparent critical period things can be accounted for by just using different learning techniques.
Some courses do try to approximate natural language acquisition to some degree. Assimil does this, but they still explain grammar, and they carefully feed you the language bit-by-bit. Destinos and French in Action have are designed around comprehensible video input, but they still ship with course materials!
This is a really superficial thing, though, and they're kind of kidding themselves on. There is nothing in nature where you pick a sequence to teach the rules then use the rules in a particular way to lead to acquisition of the features in that order.
Even so-called "natural" methods are *extremely* artificial, and if the method works, that's not evidence that the method is truly doing "like a child" stuff, but rather evidence that "like a child" doesn't work.
I actually do a lot of weird experiments in my log that try to recreate different aspects of "natural" language learning. But they're just that—experiments. And many of them rely heavily on very unnatural technology. And I'm still spending my morning setting up conjugation flashcards, because hey, sometimes that's the easiest way to keep all the forms of ser straight.
Which is great, because "aspects" is what we need to understand -- far too often people try to revolutionise teaching by throwing the baby out with the bathwater...
I think there's always going to be a Duolingo-like company in the market. Before Duolingo, Rosetta Stone seemed to hold that niche. After Duolingo, I'm pretty sure someone else will take it over. Sadly, I just don't think the market for good language courses is anywhere near as large as the market for "aspirational" language courses. But the question is whether someone can carve out the kind of niche Assimil had—enough to support a real business for many years, selling courses that actually work.
The big problem, though, is one of deflationary pressure.
When Rosetta Stone was the top of the tree, they were selling their snake oil for hundreds of dollars. It was far more expensive than any book, and they sold it by comparison to night classes and degrees, as these were more expensive. But for all of that time, there were other offerings that competed with RS on price by pegging themselves in comparison to books. There were independent courses by the likes of TeachMe, TellMeMore and EuroTalk, and then companies like Berlitz were using some 3rd party tech branded with their own name and materials. There was a course that was basically a RS clone (Livemocha) that competed on being free... so RS just bought them to make the problem go away.
But when Duolingo came out, it spurred on a new generation of language websites... that all burned through the VC money so fast that they disappeared, and even though they were better than Duolingo, only von Ahn and Hacker could get the VCs to throw in millions upon millions.