ironlady wrote:Except that a dozen exemplars are not enough for acquisition. Acquisition of structure requires literally thousands of unexpected encounters with structure that can be linked with meaning. That is the crux of the difference between rules-and-output teaching (which you seem to support: memorize and you'll "get it") and CI-based teaching. Rules and output may work for you (there are a small percentage of people who can get languages that way, and many of them become linguists or language teachers out of interest) but not for the majority.
I've got over 40,000 sentences at present, so that covers the first node, and unsurprisingly there's a lot of overlap even among the distinct sentences, even when I filter out redundant sentences. I've never studied a language in which I needed thousands of encounters with a structure to understand it and use it correctly, and neither has anyone I've ever seen study languages. Maybe acquisition of the syntax of the entire language requires that much, but there's no way a student could need as much as that figure suggests.
I've been hanging around the hostels lately. People there who take a passing interest in learning Mandarin and learn a handful of sentences, roughly parse them as they read, memorize a few extra vocabulary pieces, and try out combinations according to what they read and what their base knowledge of languages tells them (so if they know grammars from other languages, they'll approach it from those perspectives first, via rough analogical reasoning). They have a high success rate for what little vocabulary they've learned, and it didn't take thousands of unexpected encounters to get the meaning or to become familiar with the structure of their sentences. They do it like this: http://en.wikipedia.org/wiki/Example-ba ... ranslation
A dozen exemplars would be used primarily for syntactic comprehension, not overall language proficiency. Those exemplars, would, themselves, contain (perhaps) dozens of sentences, insofar as recursive languages nest simpler sentences into complex ones.
Also, I don't know what encounter would be more "unexpected" than a spaced repetition system for thousands of sentences. I certainly don't keep track of when I'm going to see sentence x again, so it's always a surprise.