To be clear, the alternative to having a pre-determined structure for language is that any sound or gesture you see/hear coming from ypur parents as a small child could potentially cary semantic meaning. Also, any characteristic of that spund or gesture (pitch, length, order, frequency, loudness, correlation with other events, time of day etc) could encode that meaning. Even once you somehow identify words as being the relevant part, you still have to identify how words relate to each other - the way order of wrods matters, the way order of sentences matters etc.

All of this ends up with a huge combinatory explosion of possible ways to assign meaning to language, requiring a huge amount of examples and counter examples to actually learn a human language. Just look at the training corpuses of something like GPT-X to see the amount of information that is required to learn elements of a human langauge if you start without any assumptions.

However, we know for a fact that human children get many orders of magnitude less examples of languahe use before learning a language - far too little to actually rule out all other possible structures if they had started with 0 assumptions.

Adding to this, there are certain elements of language structure that are not commonly thought about that happen to work identically in all natural languages that have been studied, sucha s the fact that order of words in a sentence matters, or the way certain words are omitted when constructing embedded sentences (e.g. 'can eagles that fly swim?' always having 'can' refer to 'swim', never 'fly').

It has been shown that the input that small children receive is much much larger than originally thought by the proponents of that theory. How parents stress certain words, how they make small pauses, how they look and point at things when talking, how they repeat a sentence correctly with emphasis when the child said it wrongly (Child: "apple green". Parent: "Oh, you want the GREEN APPLE? Here you are."). Wasn't there even a guy who said "How can children learn languages without pre-determined laws? Wouldn't that require being constantly corrected by their parents?". Well, see my example with the apple.

Anyway, this discussion is going nowhere. Notify me if there is other evidence apart from "I cannot imagine how the human brain can do that". As other people have written in a similar thread some months ago, this is pseudo-science.

I'd like to read more about that, could you give a source?

From my own readings, I thought that there is very good evidence adults don't bother to correct children in a lot of cases, and that children ignore the corrections completely anyway. They also don't seem to make the kinds of mistakes that you'd expect.

A great summary of some of some research I've been reading on this is the paper [1]. Also Steven Pinker's early work [2] is fascinating and easy reading.

I would love to bring myself up to date though, if these things have changed lately. Any help appreciated.

[1] Gleitman & Newport paper The Invention of Language by Children - https://www.sas.upenn.edu/~gleitman/papers/Gleitman%20&%20Ne...

[2] Steven Pinker - Why the Child Holded the Baby Rabbits - http://socsci.uci.edu/~lpearl/courses/readings/Pinker1995Ch5...

> this is pseudo-science

Couldn't agree more. It's barely even that, it's kind of just armchair philosophizing. (which is fun and often actually useful, but not sufficient to make claims about universal laws)

I would have preferred some evidence, but I guess the downvote can be also seen as an answer to my question :)

Still looking for those facts, are we?