leosmith wrote:GawainStan wrote:I can't speak for the efficacy of CI in general but I often wonder if the biggest advocates of the system are people who only ever learn 'popular' or dare I say, eurocentric languages. I have yet to find the Persian equivalent to Dreaming Spanish or a vast array of n+1
I agree with you; the "any reasonable method works" languages. Btw, I think it is i+1 rather than n+1, but I could be wrong since I've seen that before.
Let me look at a harder example than Spanish. Again; I want to make clear that everything below is experimental, and that it might not work for other people.
I was quite fond of Assimil's
L'Égyptien hiéroglyphique. This obviously uses bilingual texts, but with a 4-way alignment:
- Hieroglyphic text.
- Transliteration to a Latin-derived script.
- Literal word-for-word translation to French.
- "Smooth" translation to French.
For example:
And Assimil is more generous than usual with written explanations. You can see the ① in this image, which points to a footnote.
But still, when I was working with this, my end goal for each lesson was to cover up the two French versions of this text, and to be able to read through the Egyptian versions with at least 85% comprehension. All the fiddling around with French versions was just "scaffolding", an aid which allowed me to reach the actual goal of staring at the Egyptian and just understanding it. And to make sure I wasn't ignoring the tiny details, I made
cloze cards which hid one interesting detail at a time. This was a moderately successful experiment, although preparing the cards was laborious.
This is not a "comprehensible input method" in the sense that ALG or Dreaming Spanish would recognize. But the end goal of each little step is to produce just one more sentence of Egyptian that I can directly understand
in Egyptian. I'm basically doing what
Lingua Latina Per Se Illustrata is doing, but with a
lot more margin notes.
If you compare Assimil's course with Allen's absolutely excellent
Middle Egyptian: An Introduction to the Language and Culture of Hieroglyphs, you can see a clear difference. Allen has written a very traditional course, in fact, one of the finest I've personally seen for any language. But Allen's book is a wall of English text explaining tiny snippets of Egyptian. And if I recall correctly, it starts by explaining things like "nouns" and "verbs", because Allen couldn't assume the reader understood those ideas in English. Assimil, despite the interlinear format and the handy notes like ①, is far more focused understanding small bits of Egyptian until they add up into larger bits of Egyptian.
I actually have my favorite little diagram that tries to explain what I'm doing:
I have a
longer explanation in my log, but basically, my goal at each step is to find some trick—
any trick—which allows me to understand some bit of "input" in the target language. And then I repeat that moment of comprehension until it becomes almost automatic. The repetition may take the form of reviewing the original input, or of consuming large amounts of similar input.
I've been trying something similar for Spanish, using native audio from the beginning. My study of Spanish (which is
experimental and probably unwise) didn't start with tons of grammar study. It started with Anki audio cards like this:
Above the line, I have the "front" of the card, containing an image and a short audio clip. Below, I have the Spanish text and the English translation. I literally just started reviewing these cards "as is," with no prior study of Spanish. Which is less forgiving than
Lingua Latina Per Se Illustrata, because nobody put the input into a nice order for me, or wrote handy margin notes. But my experience was much like
Sprachprofi's earlier experiment with Japanese—within 1,000 cards and 30 hours of work, I could understand longer and longer sentences of Spanish directly from the native audio. I had transformed a small core of Spanish into an "earworm", something that was stuck constantly in my head, right down to details like the intonation. And when I encountered similar phrases in the input, I understood them automatically. This allowed me to "follow" 30–40% of the dialog. So I sat down and watched the entire series in Spanish. Which was quite enjoyable and a pleasant workout for what I
did understand. But it didn't help too much with the 60–70% I couldn't understand.
To work on active skills, I had Anki "flip" some of my cards. Here, the "front" contains an image, and some Spanish context with one phrase missing. My job is to take that English and convert it back to Spanish.
To answer these cards, I'm relying on a mix of the "earwormed" dialog, and on some primitive ability to produce Spanish on my own. Call it a 70/30 split, though it varies from card to card. This is heavily inspired by Assimil's active wave. But it's also much like
this 1922 "direct method" exercise, which starts out by telling the student
Apréndase de memoria y repítase con el libro cerrado "Learn by heart and repeat with the book closed". Which is also echoed in
"chunking" theories of language learning. There's literally nothing new under the sun; we keep remixing the same basic ideas decade after decade. Which is fine, if rather annoying for all the poor language teachers who get pushed from fad to fad by educational bureaucracy.
I do look up grammar. Occasionally. In strictly limited doses. But I have
a 6-page laminated grammar summary, and I've recently started learning the conjugation tables of common verbs using
this handy Anki deck. And the grammar study review is useful, though—like Assimil—I tend to delay it until well after I've seen the form in context. There's no reason why I can't start by watching Spanish TV, and then later circle back and learn to properly conjugate
ser.
So I hope this explains where I'm coming from here:
- I do not insist on "purity" of any sort. In my favorite diagram, the first step is "cheating." Bilingual texts, pantomime, actually looking things up—absolutely anything is fair game, if it allows me to decipher and understand some text.
- But the goal is to reach the point where I do understand stuff, so that I can "consolidate" it through sheer repetition and make it automatic.
- Output needs to be practiced separately, though there's no harm in putting it off for a couple of months. Longer than that, and it's too easy to ignore lots of interesting details that I should be "noticing."
- I do allow myself bits of grammar study. But I usually do it in a ridiculously lackadaisical fashion.
I don't actually know if this is a good method (though it seems to work surprisingly well for my goals), or more importantly, if it would actually be reproducible for anyone else. Caveat emptor. Your mileage may vary. Trained driver on a closed track: Do not attempt this stunt at home.
emk wrote:"Comprehensible input" isn't truly new. For a similar idea from well before Krashen, the
direct method was invented around 1900.
Interesting – it seems that for you, the key identifier of CIMs is that they use only L2 to teach L2. I didn’t even include that point on my list – are CIM followers these days really not using translation when consuming CI? That would surprise me, at least for the reading part. I wouldn’t consider Assimil to be a CIM, because it uses so little CI. I would consider the LingQ Method to be a CIM, even though it uses translation, because it relies heavily on tons on CI. Anyway, thanks for the links because I thought the “Nature Method” was the oldest CIM.
The "direct method" was a slightly different mix of ideas. It emphasized comprehensible input in the target language, with as little L1 as possible. But as you can see from the textbook I linked above, it also encouraged memorizing and repeating entire phrases, and of using recently learned structures to produce simple output.
But my larger point was that we've been remixing a small number of familiar ideas in simple combinations for generations now. Each becomes a popular fad, typically because it successfully addresses the weaknesses of the preceding fad. But the new fad introduces weaknesses of its own, and so the cycle continues.
How are you defining "silent period"? My definition was, except for CI consumption, not doing any learning activities. I start to pronounce/repeat stuff on day 1, but I normally wait 1-3 months before beginning to converse, so maybe we agree about the 50 days.
Well, for Spanish, I was literally watching TV with partial comprehension before I actually started properly learning my conjugation tables. My approach may not have been "pure", but it was
extremely heavily biased towards comprehensible input.
This happens with many popular methods - how many ways do people "do" Assimil, for example? That makes it hard to figure out what they mean by the Assimil method. That’s why I defined CIMs above.
This is why I wrote that big wall of text above.
I wanted to actually get into the gritty details of how I tackled Egyptian and Spanish. Egyptian is not an easy language, so any comprehension requires more "scaffolding". But the goal remained to understand actual Egyptian directly.
Cainntear wrote:But the thing is that you knew the sort of things to keep an eye out for -- you understand grammatical concepts far better than most of the average population. This gives us the weird paradox that the people who best understand written grammar explanations need them least, and the people who need grammar explanations most are likely to understand them the least.
Yes, and this is where many enthusiastic reformers fail—they discover a method which works for themselves, but which isn't actually reproducible for a broad population of students. I absolutely believe that "10,000 sentences" worked for Khatzumoto personally. But once he had paying students, he discovered that many of them were struggling massively with output. And the whole thing ended in a rather unpleasant and disreputable mess, as you mentioned.
There's a reason why my log is titled "How
not to learn Spanish"—I want to make it clear that everything I am doing is experimental, and that none of it has been reproduced with a broad range of learners. Any given step of my personal methods might be relying on some weird peculiarity of my brain. (For example, I've tried Alexander Arguelles' "shadowing" techniques, but I simply cannot hook my ears and my mouth up like that. I believe that shadowing works amazingly for him, but I need to find other techniques.)
And of course, when someone takes an interesting experimental method and tries to standardize it for schools, it's the teachers who need to clean up the resulting mess. Teachers work with a broad range of students, many of whom have gaps in their knowledge. For example, I recall the story of one poor woman who just completely failed to understand the ideas that Spanish verbs needed to be conjugated. She'd just stick random forms of verbs into her sentences—sometimes the infinitive, sometimes a different form. If I were responsible for teaching her, I would reach straight for some grammatical worksheets or some FSI drills, and I would use them to try to explain the ideas of "person" and "number". Whatever is going on with her isn't going to be fixed by piling on more input.
In fact, I'm deeply suspicious of any method that promises to get people from zero to C2 using "one weird trick". After all, you need to develop four core skills (reading, writing, listening and speaking). And the specific challenges faced at A1 are different than those faced at C2. Sometimes devoting 30 hours to a
new learning approach will produce better results than spending another 300 hours on the same thing the student has been doing all along.
All that said, I do lean heavily towards input. And not just in French and Spanish, where most any reasonable method would work. Less familiar languages require far more "scaffolding" or "cheating". But even there, sheer volume and repetition will do a lot to make a language like Middle Egyptian seem more natural and intuitive. Given a choice between Allen's traditonal and undeniably excellent
Middle Egyptian: An Introduction to the Language and Culture of Hieroglyphs, and a course like
L'Égyptien hiéroglyphique that focuses on learning to understand a series of Egyptian texts, I'm still going to go with Assimil.
But again, Assimil didn't invent the interlinear gloss! Here's
a gorgeous example from the 10th century:
The Lindisfarne Gospels (Fig. 1a) is a copy of the Holy Gospels created during the 8th century at Lindisfarne, a small island off the north-eastern coast of England. As was customary at the time, the main body of the gospels was written in Latin. In between the Latin, however, late Old English interlinear glosses were added during the 10th century by a monk and scribe by the name of Aldred, who wrote in a distinct Northumbrian dialect characteristic of the northeast area in Britain. These Old English interlinear glosses (in red) can be seen in between the Latin text (in black) on Fig. 1b. By having translated the Holy Gospels into his native Old English, Aldred provided us with the first ever English translation of the text.
And even the translation-based methods could involve surprising amounts of input. I have a copy of Gordon's
An Introduction to Old Norse, which provides the student with a short grammar, a painfully limited glossary, and a limited amount of parallel text. But the vast bulk of the book is chapter after chapter of Old Norse, which the student is presumably expected to translate. If your classical "grammar-translation" tutor expects you to translate 5 pages a day, then you're presumably going to get lots of "input"! Contrast this with textbooks that contain only a few short specimens of the target language surrounded by a wall of English text.
I am suspicious of language learning methods which never reach a point where a student is told, "Here's a book. You should be able to read this now, more or less. Go for it." Because I do think that automaticity and intuition come from sheer volume, not from reading long explanations in English. The explanations in English might sometimes be necessary, but they're a preparatory step.