|
 |
Stephen Krashen, an interview | ||
Share with:
| ||
 Language Learning Forum : Polyglots |  | |
 emk Diglot Moderator United States Joined 5320 days ago 2615 posts - 8806 votes  Speaks: English*, FrenchB2 Studies: Spanish, Ancient Egyptian Personal Language Map | Message 33 of 52 12 April 2013 at 4:48am | IP Logged | ||||||||||||||
My personal hypothesis is a bit different: The neural structures which recognize a language are different from the neural structures which produce it. You can build the recognition circuits to a very high level without building any corresponding production circuits. (Or, in the case I described, you can allow the production circuits to undergo near-complete atrophy while still maintaining the recognition circuits.) When you "activate" a passive language, I suspect that the production circuits start by spewing all sorts of semi-coherent babble. The recognition circuits watch, and provide rapid feedback: "Seriously. That sounds horrible. Nobody speaks like that." If this were true, it would explain why people can have strong comprehension but weak production. And this describes quite a lot of bilingual preschoolers I know—they can follow complicated instructions in two languages, but they only speak the language they actually need to speak. So no matter how good your passive skills get, if you don't actually try to communicate (at least via talking to yourself), there's no way to use that passive knowledge to tune up the production circuits. Or at least this seems to be true for a significant fraction of the population. Without a need to speak and to be understood, people of all ages routinely fail to develop active skills.
Are you speaking about Error-Related Negativity? (A brain signal also amusingly called the "oh shit wave".) There's certainly evidence that ERN is involved in recognizing language-related errors. There are also signals which fire for grammatically broken input, and there's no reason why those circuits couldn't be used to monitor output as well. The interesting thing about ERN is that it's very general—it can be triggered by consciously-perceived errors, but also by errors that never cross the threshold of consciousness. It would be really surprising if the language production centers couldn't accept ERN as feedback.
Yes, I know feeling well. :-) It think this is one of the major ways in which conscious knowledge can provide a huge boost to language learners—it doesn't matter why input is comprehensible, only that it's understood. This is why I spend a lot time cheating to boost my comprehension.
That's a great example of what I mean. A different example might be the "learn something, see it everywhere" effect. I know I'm not the only person who can read a grammar lesson on the imperfect subjunctive and see it leaping out at me everywhere I go for the next several days. Presumably it was there all along, but by learning it consciously, it suddenly became highly noticeable, and I got a big surge of comprehensible input. The same thing applies with informal grammar—some pattern will catch my attention, I'll notice lots of examples, and I'll chew on them a bit without formulating an explicit theory. I think most of our cognitive access to our language-learning machinery relies on various sorts of tricks like these. But I've spent too much time watching kids learn to speak to believe that everything is subconscious—even 2 year olds will sometimes slow down and think before putting a longer sentence together, and it's adorable when they suddenly enunciate each word very clearly. They're obviously not using explicit grammar rules, but they're clearly thinking about what they're saying on some level. 3 persons have voted this message useful | |||||||||||||||
| Volte Tetraglot Senior Member Switzerland Joined 6227 days ago 4474 posts - 6726 votes Speaks: English*, Esperanto, German, Italian Studies: French, Finnish, Mandarin, Japanese | Message 34 of 52 12 April 2013 at 5:20am | IP Logged | ||||||||||||||
|
1) It is - and even adults can be taught to read.
2) It wasn't an argument against Krashen; I think his core idea is largely correct. It's an argument against some formulations of a LAD. 3) Help, yes. Get me all the way there, no. Müller-Lyer took significant time to consciously adapt to. 45 minutes isn't sufficing for the checkershadow illusion, and I'd rather not spend all day on it. It's sufficed to change my perception of it fairly drastically, but I'm still not managing to see A and B as the same color in an unaltered version of it, though I do see more places where the surrounding 'squares' have been carefully altered to enhance the illusion. In altered versions of it (with a small rectangle of the same color between the squares), I can sometimes make them appear the same color at this point, but it's a rather tiring mental effort and tends to get processed as a gradient, which is quite hard to suppress. 1 person has voted this message useful | |||||||||||||||
 Iversen Super Polyglot Moderator Denmark berejst.dk Joined 6491 days ago 9078 posts - 16473 votes Speaks: Danish*, French, English, German, Italian, Spanish, Portuguese, Dutch, Swedish, Esperanto, Romanian, Catalan Studies: Afrikaans, Greek, Norwegian, Russian, Serbian, Icelandic, Latin, Irish, Lowland Scots, Indonesian, Polish, Croatian Personal Language Map | Message 35 of 52 12 April 2013 at 11:24am | IP Logged | ||||||||||||||
|
The idea about comprehensible input didn't originate with Krashen, but he was the one who formulated and popularized it, and for that alone he deserves to be seen as one of the great language acquisition thinkers. But as all reformators he also incorporated some dubious elements in his theories, and I can see from my own language learning that the factors which decide how well I an speak and write a certain language don't follow the lines laid down by Krashen.
For instance I can see that my progress with specific constructions is closely related to my conscious knowledge about those structures. There will be a phase where I construct my sentences rather than utter them freely, but that's a just a phase. And much depend on the way the grammatical knowledge is treated. For instance I use my green sheets as visual roadmaps to morphology, and I use rules of thumb to guide me in cases where I have to formulate something and it just doesn't materialize by itself. But that grammatical knowledge has to be formulated in operational terms. For instance I tend to remember morphology as locations on my green sheets, and that can mean that I sometimes have started to write or even say a sentence, I know that I need a dative in a moment and then a picture of a green sheets flashes through my mind and the endings pop up ind my mind. Or I imagine the structure of a 'sentence knot' as a system og boxes and sticks even while I'm making it up. So I have to answer 'yes' to Emk's 3. question: "Can approximately correct intellectual knowledge be used to bootstrap the acquisition of detailed subconscious knowledge?" I can simply see that process happening in my weaker languages before the mechanisms become automatized and invisible. Would it work without that intermediary conscious step? Well maybe, but not without a lot of input AND surroundings where I had to produce utterances myself. And that points to the reasons why I still have languages which I have studied for years below the 'speaks' line. The key is the notion of comprehensive input. Only comprehensible input (spoken as well as written) is relevant for language learning. But the definition of comprehensible input can be softened up by using input where you already know the meaning, and that's the reason that the use of bilingual texts is so efficient in language learning. In the video discussion between Krashen and Steve Kaufmann the latter also said that he could tolerate up to 40 % unknown words in a text because he could get them translated fast through his software. The turning point for me comes when I can sit down with a text and read it extensively or listen extensively to a TV program - maybe with subtitles, if it helps. Only with enough comprehensible input is it possible to get through so much input that your head starts buzzing with fragments of the foreign language - otherwise you'll be bored to death. Language production before that buzzing stage will consist of a mixture of prelearned phrases and slow, cumbersome constructions based on rules and isolated elements. Or in other words: the buzz which comes from an overdose of comprehensible input is for me the prerequisite for learning to think, write and speak spontanously. But the buzz only works because most of the gramamtical structures and words already have been learnt beforehand, otherwise that input wouldn't be comprehensible. But how you get the structures and the words that make the input comprehensible is quite another matter, and as far as I can see it depends on yourself, on the amount of exposure you can get and on the level you are at in your target language(s) which methods you should use. For me at least the formal (and conscious) methods function better than the theories of Krashen allow for. PS: my answers to EMK's two first questions: 1) Can the brain accept explicit examples of acceptable and/or unacceptable sentences and learn rules from that? What about rehearsing acceptable sentences? No need to teach grammar, just tag certain data points for special interest. Maybe if the examples are presented in a structured fashion where something is illustrated by sentences with just one difference. And that's a rare case. Or if you get so much input that you see the same elements recurring time and time again - then you may notice the regularities. But it takes time, and knowing what to look for would make that process much more efficient. 2) Can you create comprehensible input the hard way, using either translations or explicit knowledge of grammar to understand something? What if you re-expose yourself to the now-comprehensible input on a regular basis? Certainly. I make bilingual texts galore, and I have found that it helps even when I can read most of the text without help. Grammar has a slightly different role: it is not easy to look things up, but if you have read about some phenomenon you will see it pop up everywhere. If you don't then learning that part of the grammar isn't essential. Edited by Iversen on 12 April 2013 at 12:26pm 4 persons have voted this message useful | |||||||||||||||
| Serpent Octoglot Senior Member Russian Federation serpent-849.livejour Joined 6385 days ago 9753 posts - 15779 votes 4 sounds Speaks: Russian*, English, FinnishC1, Latin, German, Italian, Spanish, Portuguese Studies: Danish, Romanian, Polish, Belarusian, Ukrainian, Croatian, Slovenian, Catalan, Czech, Galician, Dutch, Swedish | Message 36 of 52 12 April 2013 at 2:34pm | IP Logged | ||||||||||||||
|
A couple of obligatory links I can't help posting:)
2 persons have voted this message useful | |||||||||||||||
| patrickwilken Senior Member Germany radiant-flux.net Joined 4321 days ago 1546 posts - 3200 votes Studies: German | Message 37 of 52 12 April 2013 at 2:48pm | IP Logged | ||||||||||||||
Of course trivially that must be true. There are neural structures that deal with muscles etc involved in language production that are separate from whatever grammar circuits there are. But since it doesn't follow that these language production circuits understand explicit grammar rules any better, your friend's example doesn't seem to me to argue against Kreshen. I find your language learning interesting as there are a lot of parallels between us. I am married to a German, tried for years to learn on and off, finally started self-learning nine months ago, partly out of fear that I won't be able to speak to any future offspring, partly so I can work more easily here in Berlin. I am mostly doing lots of input (movies and books and anki). The big difference between us is that I am not trying to talk too much. I am not against talking, it's just that my wife and I are so comfortable in English it's almost impossible to switch over. Most of the conversations I am having are with random strangers when I go shopping or at bars (at most one hour per week - as opposed to say 20-30 hours/week of input). What's interesting is that on the rare occasions I speak everyone is amazed at how much better my spoken German is getting. It's not perfect, and it's definitely lagging behind my passive knowledge, but it's getting better all the time without really any work. It's not that I don't think I'll ever have to practice speaking, but at the moment my progress is fast enough that I am not going to worry about it too much. When I get better (say C1 in comprehension) I am going to start writing more and speaking more and see what I need improve. This isn't a pure experiment, but I will be curious to see what the strengths and weaknesses are in six months. It comes down to what you think the passive vs active distinction in language is. My guess is that passive and active knowledge are the same circuits, it's just that passive is less well established, so it's less likely to be spontaneously activated, which fits with the way I can understand words in context reading, but not actively produce them in speech. Also I think that your internal grammar can be quite simple and still allow you to understand a sentence. If I give you a three word sentence which contains (cat, mouse, eat) you can probably guess who is eating whom, without needing to have in your brain the correct grammar rules for word order, let alone necessarily have encoded the the correct way to decline nouns, adjectives, gender etc. It's not surprising to me that I can read fairly comfortably, but not be able to generate correct case/gender etc in spontaneous speech. In this view what you need for correct spontaneous speech is better circuits, not practice with output. However, it might well be that to fine (even coarse) tune those language circuits you need to get feedback about errors you are making (see below). So in that sense talking/writing might be critical to learning.
I was thinking more generally in a neural net sort of way, but yes, exactly perhaps the ERN when we recognize that we make a mistake is helpful in learning, which if it were would argue against the simple idea that input is the only useful thing. It seems very plausible to me, but I guess it's an empirical question. One thing that might be interesting is how close in time the ERN (or whatever error signal you have) has to be to the output. It might be that you need a fairly tight time coupling, so if you have explicit knowledge and can correct your own errors on the spot, you might learn a lot faster (than say someone pointing out these errors later). It occurs to me that shadowing might be useful precisely because you are matching your output in real time to input and so able to see and correct errors pronunciation etc in real time. ----------- Just as an aside to another post: it's well established in a lot of other physical activities that learning starts slowly and consciously and then becomes fast and unconscious with practice. I remember very clearly how conscious and clumsy I was when I first started riding a bike or when I first learnt to touch-type. Now if you asked me where the letter 'p' on the keyboard is, I couldn't tell you consciously, what I do is move my little right finger up so see where it is. It might well be that language learning starts consciously (even with grammar rules etc) and then goes unconscious after much practice. ------- Serpent's link to a story about someone who learnt to communicate without speaking/writing does sort of argue against output being that important: http://www.antimoon.com/how/input-boydell.htm Edited by patrickwilken on 12 April 2013 at 3:05pm 1 person has voted this message useful | |||||||||||||||
| DaraghM Diglot Senior Member Ireland Joined 5939 days ago 1947 posts - 2923 votes Speaks: English*, Spanish Studies: French, Russian, Hungarian | Message 38 of 52 16 April 2013 at 10:55am | IP Logged | ||||||||||||||
Isn't this the essence of skills acquisition ? I used to believe in language related areas of the brain for L2 learning, but I now think this only applies to your L1. I think L2 acquisition is in the same domain as skills mastery and also requires motivation, extensive practice and feedback. 1 person has voted this message useful | |||||||||||||||
| patrickwilken Senior Member Germany radiant-flux.net Joined 4321 days ago 1546 posts - 3200 votes Studies: German | Message 39 of 52 16 April 2013 at 12:11pm | IP Logged | ||||||||||||||
That's my intuition too, but if that's correct than Krashen is wrong at least in part. I agree with EMK that Krashen is about 80% correct, in that lots of input is needed, but that's not the whole story, and error feedback is also important. That doesn't mean that the same areas that are active in learning an L2 are different from learning an L1 though. I think children do just as much work learning a language as adults. Probably more: I read yesterday in passing in a paper from the Foreign Service Institute that a child spends about 15000 hours learning their L1. 1 person has voted this message useful | |||||||||||||||
| emk Diglot Moderator United States Joined 5320 days ago 2615 posts - 8806 votes Speaks: English*, FrenchB2 Studies: Spanish, Ancient Egyptian Personal Language Map | Message 40 of 52 16 April 2013 at 1:10pm | IP Logged | ||||||||||||||
There are several studies suggesting that advanced L2 speakers use pretty much the same parts of their brain as native speakers. For example:
I would agree that many beginner and intermediate learners rely on "skills mastery" and practice. I don't, especially—I mostly just read and listen and talk to people, and spend a small amount of time reading about grammar. And as demonstrated by the study above, I think this distinction largely disappears for most adults once they reach C1. I know a huge number of grad students, post docs and faculty who've made it from C1-ish levels to near-native levels without any explicit study at all. I've noticed couple of big differences between kids and the typical intermediate L2 learner: 1. Preschool kids get 3 to 13 million words of input per year, much of it comprehensible. This is about 12/hours per day for monolingual kids. 2. Kids spend much of the time between age 2 and 5 learning to speak. 3. Kids are forced to interact with their parents and other kids. 4. Kids seem to have strong listening comprehension, relative to their output skills, compared to most adult learners. This makes sense: They can't learn something until they're accurately heard it plenty of times. Most adults try to master the subjunctive before they can actually watch TV for fun. And as patrickwilken mentioned, kids work really, really hard to learn their L1. And if they need to pick up an L2 during their childhood, it's still a lot of painful work:
If there's a difference between childhood acquisition and adult acquisition, it may lie in the fact that only about 5–10% of native speakers need feedback from a speech therapist to clear up various weird problems in a timely fashion, but that a much larger fraction of adults would benefit from similar feedback from a language tutor. Edited by emk on 16 April 2013 at 1:12pm 3 persons have voted this message useful |
|  | 
Printable version
|
| Copyright 2024 FX Micheloud - All rights reserved No part of this website may be copied by any means without my written authorization. |
