Neurolinguistics (Giosuè Baggio) wrote:We also know that speech processing----like virtually any other perceptual and cognitive process in the brain----is predictive: the brain is never waiting passively for new input to arrive; it is always trying to anticipate the next stimulus in a sequence and the outcomes of the organism's own acts. Moreover, there are different kinds of prediction, which may engage different predictive systems in the brain. Predictions may rely on knowledge of regular phonological properties of speech or, equivalently, acquired phonological patterns in a given language, which would make certain sequences of sounds more likely to occur, that is, more predictable, than others. Based on this prior knowledge, as speech unfolds in time, the brain actively generates a model of the current stimulus and its possible continuations.These internally synthesized representations of spaeech are then compared with the actual input, and the error (the difference between the internal model and the input) is fed back to the system for online correction.
I don't know whether this is something I was told before or it just matches my intuitive understanding... but I suspect both. I'm pretty certain that in my distance degree module covering linguistic principles and the grammar, the concept of "garden path sentences" was used to illustrate the idea that language was predictive, and jokes can be more funny if they lead you to expect something and surprise you with something else -- the name "garden path sentences" references the idiomatic phrase "to lead someone up the garden path" describing exactly that. The common example used was a Groucho Marx line: "time flies line an arrow; fruit flies like a banana." His first clause is a simile and set up a structure of #(noun) flies (verb) like (preposition) a #(noun) that we expect to repeat -- our brains see the second clause as a parallel and it's only when the final noun appears that the semantics break down. (And actually, I think he cheated by saying "a banana" rather than "bananas", because "fruit flies like a banana" is potentially not valid English, but never mind... it still illustrates the point!!!)
So that's a major underlying part of my disagreement with Krashen, because regardless of when I first read about predictive stuff, I had always felt that I didn't fully understand something in a foreign language if I couldn't (theoretically) say the same thing myself. If I didn't know the grammatical structure, I'd be basically guessing at the intended meaning. I say "theoretically", because the real edge of my language development were sentences that I couldn't say spontaneously, but were made up of a combination of features that I could use spontaneously.
But if comprehension relies on being able to anticipate likely next words, how can comprehensible input even exist? That's why I think that Krashen's constant use of German as a demonstration lesson was something of a cheat, a get-out. German is just similar enough to English that you can start off with pretty reasonable predictions without being too far wrong that you can't pick it up in the "correction" phase. French would probably be close enough too. But if he'd done his demo with Basque... what would people have thought?
Hau da nire eskua.
Bi esku ditut.
Hauek dira nire begiak.
Hiru begi ditut: bat, bi, hiru.
"Ich habe" is close enough to "I have" to be assumed to be the same... and assumed correctly. "ditut" is pretty far from "I have", but the assumption is still fairly likely... but Basque is so different that the assumption is wrong.