The use of FSI, a question of efficiency.

[Tomás]

OP may want to have a look at:

http://www.dicendipublishing.com/

It's an Iphone course based on FSI Basic Spanish. I think it looks pretty great, and I would purchase it if I had Iphone.

My major beef with FSI is that it is unwieldy to use. The lessons are long, and sometimes you need to consult the book. This app breaks it up into bite-size chunks that you can use independently of the book. I wish he'd build one for Android.

[kunsttyv]

OP may want to have a look at:

http://www.dicendipublishing.com/

Even if you don't want to buy their ibooks, I find the unit index very useful, to see a list of grammatical topics covered in each unit.

I haven't even opened the course PDFs, I use FSI as a pure audio course, and before each unit I work my way through the grammatical topics of the day (with GdUdE). It's not hard to see how for some this course would be the definition of drudgery. For me, it's something enjoyable and I suspect really useful for my Spanish skills.

The course storyline is actually quite hilarious considering the ridiculous reality and lifestyle of these 1960's diplomats. For me, it works like a funny and slightly unreal kind of first hand historical narrative.

[William Camden]

Then again, Luxembourgers are right at the frontier between French and German. I think people learned and still learn well by grammar-translation if they want to and try hard enough. The same with just sucking up the world around them.

The single most important part of the linguistic abilities of Luxembourgians is their education system. I don't how or when it was established like that, but they have their lessons in secondary school in German, English, French and Luxembourgish, in some schools also Portuguese. They might have biology in French and history in Luxembourgish, that sort of thing. A lot of immersion. At least that's what my Luxembourgish friends told me. I've met plenty of them in Germany and England and they are pretty much the only folks who ever fooled me with their German abilities. They speak pretty much without an accent - very minor if at all -, but sometimes they use very strange and archaic words or slip up in some other very hard to recognise way.

This is a little at a tangent, but about ten years ago I was in Brussels and speaking French with Belgians. I was asked if I was from Luxembourg. Apparently my French was correct but my accent had a slightly Germanic tinge that they associated with Luxembourg. Or it might be that a slight Scottish accent was mistaken for Luxembourgeois (when I speak L2s I make an attempt to lose my native accent but I don't think I succeed, at least not 100%).

[Cainntear]

Whenever I come across bold statements such as, let's say, that the audio-lingual methodology has been debunked by modern theory on second language acquisition, or that FSI courses are by nature inefficient, when it's obvious that this is not the case for the many people who have had success using these methods as part of their language learning routine, I tend to just filter out those voices from there on.

First up, there's a big difference between success and efficiency. You can successfully light a fire with a few sticks, but a box of matches and a firelighter are going to do the job quicker. So if some of us say that FSI isn't efficient, we're not claiming no-one can learn from it.

Secondly, there's a general observation that some people will learn whatever the technique. Judging a technique's effectiveness by the performance of the top of the class is worthless, as the top of the class would almost certainly do really well with a completely different technique. (Please note that I'm not talking about some idea of "inherent intelligence" here -- the cause might be previous education, home background, self-confidence, whatever. But the top of the class are in some sense "good students".) The (often forgotten) rule of thumb in school teaching is that you should look at the middle third of any class if you really want to see what difference the teaching has made. This is obviously harder to do where courses have a hig drop-out rate, and self-access materials in general do have a high drop-out rate. But still, judging effectiveness on a handful who complete the courses is judging the course by the performance of the top students, and like I said, I don't think that's valid. (I'll also point out at this point that the FSI themselves are highly selective with very competitive entrance exams, so practically their entire class is the top of the class -- and besides, the materials we have access to are a tiny part of the course,)

Now when I talk about behaviorist audio-lingualism being disproven, I'm not saying that there aren't elements within the teaching that are good -- all teachers blend techniques from experience with theoretically or ideologically motivated techniques; that's why people have historically managed to learn despite all the new and latest theories. There was a link to an early video of the "Army Language School" posted here a couple of months ago, and what they showed you from inside wasn't really audio-lingual at all. In general I find that what people put down on paper follows theory much closer than what they do that isn't written down (e.g. delivering live classes).

This is especially true when the bold statements are accompanied by some personal sets of theoretical underpinnings that are meant to explain (scientifically) why the others have got it wrong.

Here's something I don't think I've said for a while:
Most people who fail fail because of the materials; most people who succeed succeed despite the materials.

If the material presents two ways of completing the task, but only one results in learning language, then you would expect on average half of the users to learn the language, all else being equal.
The possibility of completing tasks successfully using incorrect techniques is a weakness in any course, and the goal of the course designer is to eliminate false paths and dead ends.

Going back to RS (a less controversial case than FSI), it's perfectly possible to start associating full sentences with pictures without thinking about the meaning. The task doesn't force you to deal with it as a piece of language. If you brought 100 people to my door who had learned really well with RS, I would still assert that lots more failed, and that their failure was because RS's core design philosophy doesn't immediately reward doing things the right way, and by the time the delayed reward shows up, it's just too late to change the way you work with it.

Of course, Rosetta Stone's design philosophy was weak because it wasn't written by people who knew anything about teaching, but FSI was written by people who did know about teaching (so contains some good things) but on flawed theoretical grounds. The audio-lingual method simply does not require you to think about meaning. People who are successful with FSI presumably force themselves to do so -- I know that when I was at high school we did a lot of drill-based learning, and I put a lot of effort into trying to draw meaning from all the tasks, rather than following the tasks mechanically; but you shouldn't have to force yourself to do that -- the materials should always reward doing it right.

[jsega]

you shouldn't have to force yourself to do that -- the materials should always reward doing it right.

Would you mind sharing some examples of materials that do this right and how?

[Cainntear]

you shouldn't have to force yourself to do that -- the materials should always reward doing it right.

Would you mind sharing some examples of materials that do this right and how?

Ah yes, the Holy Grail.

Right, well., here's a scale of bad to good.

I have in front of me a book called Spanish Verb Drills. I seem to have written in about 1 page of it in total, because I didn't see and value in it.
It's based on mechanical drills of several styles, including:
* Presents a pronoun and an infinitive verb, and asks you to conjugate it to one or more tenses by name. As these are regular verbs, you can work by the ending without thinking about what the root means.
* Write out a full conjugation. As conjugation tables have a fixed order, you don't need to think about the person to get the right answers.
*Translate fragments from English to Spanish. The fragments really don't mean anything at all. e.g. "We were loosening." Until you tell me what was being loosened, there is no meaning whatsoever in there, so you can't think about meaning.

I used to have another book called Practice Makes Perfect: Spanish Verb Tenses. They had some similar exercises, but they built up to translations of full sentences, so unlike with the other book, you were doing something to something. Because the prompts had clear meaning, you started to think about meaning. Furthermore, it meant that the prompts following each other were quite different from each other, and the more each individual task is different from the last, the more you have to think to get the right answer, rather than just holding what you need in working memory and only changing a tiny bit of it. However, the weakness was that the one part of the sentence you didn't have to think about was the tense, as that was the bit that didn't change from one task to the next, and that was supposed to be what you were practising. However, there were always mixed revision tasks at the end of sections, so it wasn't a fatal flaw.

Then we get to Michel Thomas, which is my good end of the scale. Thomas doesn't do a perfect job, but one thing you'll find is that if you look at consecutive prompts, they will only be similar when he's dealing with something that is genuinely new. He strikes a very fine balance between allowing sufficient overlap and link in meaning that there's some support for the next and keeping things different enough that you have to think through the whole thing. And crucially, he makes sentence length part of this. The better you know something, the more complex the setting you're going to be expected to use it in.

When it comes to trying to force people to work linguistically rather than mechanistically, complexity is the most effective tool -- mechanistic production from conscious rules places a heavy load on working memory, and a long sentence will overload working memory, so using linguistic processing is the only option open to you.

[blaurebell]

I know that when I was at high school we did a lot of drill-based learning, and I put a lot of effort into trying to draw meaning from all the tasks, rather than following the tasks mechanically; but you shouldn't have to force yourself to do that -- the materials should always reward doing it right.

I'm not sure meaning is really that significant for the effectiveness of the drills. The only language where I've ever spent time on seemingly pointless mechanical drills without caring about it at all is English - those were mandatory English classes in German high school and I just couldn't wriggle my way out of doing those damn boring grammar drills. I've done surely about 4 or 5 whole books of grammar drills during high school. English is also the only one of my languages where I have a good grasp on basically all of the grammar and have complete automaticity on all aspects of it. I probably know more about English grammar than about German grammar actually.

Maybe we should just be more precise what we mean by efficiency here:
1. FSI is an inefficient method to get to a good level of comprehension, because it focuses on production and not on recognition of grammar. There is almost no need for drilling to recognise grammar patterns. Especially with some languages you will need to have reached the end of FSI to have seen enough grammar to be in a position to understand even fairly simple children's books and you might still not have enough vocabulary for them. And of course you will probably drop the course because you will see absolutely no improvement in comprehension until you reach the end and it's absolute torture in the meantime. To jumpstart comprehension I find Assimil + Duolingo + intensive reading the most efficient and least painful approach.
2. FSI is an efficient method to get to good levels of grammatically correct production, at least according to everyone who has actually finished the whole thing. Oodles of correct input might get you there too eventually, but it will take substantially longer than 90h! More like 9000h. In Spanish I have had about 350h of not too grammar heavy language classes - very few grammar drills, because you don't torture paying customers. I have read thousands of pages with different difficulty levels and genres. Make that another 300h at least. I have also watched hundreds and hundreds of hours of TV - probably around 500h-700h. I have even heard Spanish spoken every single day on the street for 2 years now. That's 1000h of input at least +classes. And I still make mistakes with fairly common subjunctives!!! It drives me nuts! I know when I make a mistake thanks to all the input, but I still make the damn mistakes anyway, probably due to some deeply ingrained bad habits from starting to speak too early. Next year I'll try to do FSI Spanish. Maybe that will finally sort it out

[Cainntear]

Maybe we should just be more precise what we mean by efficiency here:
1. FSI is an inefficient method to get to a good level of comprehension, because it focuses on production and not on recognition of grammar. There is almost no need for drilling to recognise grammar patterns. Especially with some languages you will need to have reached the end of FSI to have seen enough grammar to be in a position to understand even fairly simple children's books and you might still not have enough vocabulary for them. And of course you will probably drop the course because you will see absolutely no improvement in comprehension until you reach the end and it's absolute torture in the meantime. To jumpstart comprehension I find Assimil + Duolingo + intensive reading the most efficient and least painful approach.
2. FSI is an efficient method to get to good levels of grammatically correct production, at least according to everyone who has actually finished the whole thing. Oodles of correct input might get you there too eventually, but it will take substantially longer than 90h! More like 9000h. In Spanish I have had about 350h of not too grammar heavy language classes - very few grammar drills, because you don't torture paying customers. I have read thousands of pages with different difficulty levels and genres. Make that another 300h at least. I have also watched hundreds and hundreds of hours of TV - probably around 500h-700h. I have even heard Spanish spoken every single day on the street for 2 years now. That's 1000h of input at least +classes. And I still make mistakes with fairly common subjunctives!!! It drives me nuts! I know when I make a mistake thanks to all the input, but I still make the damn mistakes anyway, probably due to some deeply ingrained bad habits from starting to speak too early. Next year I'll try to do FSI Spanish. Maybe that will finally sort it out

What you've presented here is two extreme viewpoints and presented them as natural alternatives to each other, but there are lots of midpoints you've left out. "Oodles of correct input" is not the only alternative to oodles of mechanical drills.

The history of language methods is one of overcompensation -- one course does too much of one thing, so the next does none of that and does too much of something else instead.

I've met lots of people who've gone through grammar-heavy courses and can do all the tasks, but fail in real production because while they can produce the grammar on demand from the teacher/book/CD/website, they can't do it when they're speaking. What is the problem here if it is not the failure to link the form with the meaning?

I learned my Spanish initially in a mostly production-based way, and my comprehension came as a natural consequence, because I knew both the form and the meaning of everything I said in practice.

[Systematiker]

I'll chime in here:

My Spanish is pretty unbalanced. I watch film and listen to podcasts and lectures without missing anything. I read novels and nonfiction in my fields as well as areas of interest and I can measure unknown words per 100 pages and still be in the single digits. I can write, at least in text messages, pretty well, though not as well as that level of comprehension might lead one to expect.

And in the last three weeks, I've had two drop-of-the-hat conversations in Spanish where I'm lucky if I expressed myself at B2.

So I got irritated about this yesterday, and I've begun a process by which by the end of March I will have shadowed every dialogue and done every oral drill in the whole of the FSI Spanish program by the end of the month. I'm not learning anything new - I know all these words, and I know all this grammar, at least in theory. But it damn well better automate my production. I think (and hope!) that Blaurebell's #2 is correct. I'll find out.

[jsega]

What is the problem here if it is not the failure to link the form with the meaning?

I learned my Spanish initially in a mostly production-based way, and my comprehension came as a natural consequence, because I knew both the form and the meaning of everything I said in practice.

What do you say to those who use the idea of "muscle memory" as an explanation to this?

Martial arts training seems to be a popular analogy for describing speech production as a muscle (or network of muscles) that needs forms to be drilled in order to expect reliable and automatic execution when it counts.

If this is a valid point, it would seem FSI-type drills would be ideal for the serious language learner.