Is consuming material that's beyond your "comprehensible input" level ever BETTER than understanding what you consume?

[Cainntear]

The "standard" advice I've heard given by schools and language teachers is "no more than 5 unknown words per page."

But the reality is that it's really hard to find books and media that are that close to my level. I have to work with what I can get, and until several years ago, that meant ordering DVDs from France and hoping they had accurate French subtitles. And then buying a region-free drive to watch them. So I don't think I ever got the luxury of following the "no more than 5 unknown words per page" advice during the B1 and B2 levels.

FIVE WORDS PER PAGE? That's just astounding to me--there've been many times when I could more or less understand a narrative despite there being far more than five new words per page. What's relevant, IME, is not so much how many of them there are, but how bunched-up they are. If you don't understand five words in a 10-word sentence, then you likely won't understand the sentence at all. But if you have 15 new words spread more-or-less evenly over a page, there's a good chance you can learn them from context.

I think the whole thing comes from well-meaning attempts to answer questions about what "Comprehensible Input" really means, because it's something that has been so vaguely defined for so long. 5 is by itself a suspiciously round number, but I've seen criticism that teachers were told to teach "no more than 7 words per lesson", and although I've only seen it as criticism and not actual advice, I imagine it's coming from an overinterpretation of "the psychologist's magic number": 5 plus or minus 2. That was what I remember being told described working memory -- the claim was that the average person can process between 3 and 7 things in working memory and can only identify that many distinct items in view without actually counting.

This is where chunking comes into things, because the idea is that by grouping a bunch of concepts together and making the collection into its own concept (like "shot ...self/selves in the foot") we reduce the load on working memory from a notional 2 or 3 down to 1.

It looks like they may have revised it down to 4 (or 3 to 5), which I'm happy about because I was aware that I was failing to recognise more than four things visually -- I could see that a normal bass was a bass because it had 4 strings, but I could only see that a normal 6-string guitar was a normal 6-string guitar because it had a certain symmetry -- I could split in half and see three strings and three strings. With a 5 string bass, I would see that there was a string serving as the axis of symmetry, where a guitar had the axis in a gap.

So the old logic behind "no more than 7" seems suspiciously similar to the logic behind "no more than 5".

The problem is, though, that working memory is a very short term thing -- you will be using working memory plenty of times during a lesson of even half-an-hour or on a single page on a book, and even if you're not starting from zero every time. there's not really any reason to see a page as a boundary on working memory. It's a pretty arbitrary rule-of-thumb, and it's only useful if it's recognised as such, and individual practicioners know that they should apply judgement... but like so many things, "what gets measured gets managed" and any attempt to apply judgement risks getting clocked in a performance review as "going over the recommended limit."

Enough repetition of material that's far above your level can hammer it in--I've learned that for sure. It's just a question of whether it's better to spend the time hammering, or to read easier stuff that doesn't require it. I suppose you could hedge your bets by training both ways.

Well, I think the difference with the advice under discussion is that teachers would be looking at time constraints. You don't want your students to be set indefinite tasks -- they need to be in control of time. Individual learners working without a teacher can spend as long as they want to work out what the page says, but in a class, people who flash through it will finish the passage before the slow and methodical ones have even found their first unknown word in the dictionary. Whatever you do, you're going to lose some of them. Even if you're giving the reading as homework, you'll end up with the methodical ones feeling overwhelmed, and there will be less time-on-task for the quick ones, so potentially less learning anyway.

And going back to the working memory numbers, the "no more than 7" and "no more than 5" rules are ignoring the fact that 7 and 5 were proposed *upper limits* for average human performance; which means most average humans fell under that. Is it really a good thing for a teacher to only teach for the absolute best in the class...?

[garyb]

This just sounds like the age-old discussion of intensive verus extensive work, which of course is a spectrum rather than two distinct things so it's hard to define what each one really is, although that doesn't stop people with time on their hands from trying and arguing over it (so hoppefully I'm not doing the same here by adding my definitions!). But it seems like the norm is to do a mix of focused, slow work on small pieces of language that might have a lot of unknowns ("intensive") and faster consumption of more language that has fewer unknowns ("extensive"), and they're both useful and complementary activities. I don't think you can really compare them and say that one is "better" than the other, although one might be more productive than the other at a given time depending on level, goals, and weak points.

[Cainntear]

Can I just ask you what your definition of comprehensible input is? Because if you have to do all that to understand a paragraph (or a page or a chapter), then I wouldn't say it was particularly comprehensible for you.

To be fair, he started this thread to discuss whether reading stuff that's harder than "comprehensible input" was useful, so he wasn't really talking about CI.

I think the main thing here is that CI is defined as something you can understand without looking anything up. Anything where you have to look things up isn't CI, and the Krashen line is that this falls into learning about language rather than straight up learning language (he uses the distinction of "learning" vs "acquisition", but that just seems arbitrary semantics to me, and most of the SLA world doesn't make that distinction.)

The underlying point behind Khayyam's posts is that he doesn't believe in Krashen's "acquisition-learning distinction" and believes that working on something so hard that you have to actively think through it and look stuff up is good.

I agree.

I think the problem comes whenever people look for the *only* thing to do and then do it to death. Doing too much "intensive reading" isn't going to lead to fluent reading, and doing a lot of reading without intensity (e.g. comprehensible input) isn't going to lead to accurate reading. Fluency and accuracy are both important.

[jeffers]

I would say variety is the way to go. Read or listen to some things at your level, some things above your level and some things you find really easy. Each has separate benefits. Briefly, reading things which are "too easy" helps you to get comfortable with standard structures without having to be distracted by wondering what things mean. Reading things which are "just right" (e.g. a few unknown words or structures) is good because you can work out those unknown parts by context, or look up the one or two you can't, so you are gently building your vocabulary. But working on things which are quite a bit above your level can be useful as well, but I would say mainly as listening practice, because you get to know the rhythm and feel of full speed native language. The balance is really up to the individual, their tastes and interests.

Here's one example from my experience. When I had been studying French for around a year I bought the book and audiobook of Le petit Nicolas, intending to study the written text intensively before listening. But simply because I was taking long walks on the weekend, I listened through the whole book 4-5 times before I even looked at the text. The first lime through I understood very little, but I got a sense for what the stories were like and I was able to pick out a few words here and there. Each subsequent listen gave me a lot more information about each story and my comprehension grew each time. When I finally read the book I started each story with a good idea of what it was about and where it was leading, which made it easier to read. My practice ever since, with shorter texts that have audio, has been to give it a few listens first before I actually read the text. Sometimes after a couple of listens I don't feel the need to read the text, other times I might just check a few sections.

Regarding the "5 unknown words per page" thing, I think this is probably based on Paul Nation, who says that a reader should have 97%-98% coverage of the words to "read with pleasure", which is quite a different thing from doing an intensive study of a text. Actually, 97% coverage would result in one unknown word every 3.3 lines of text on average from his corpus, so that's going to be a lot more than 5 per page. I haven't read this article in full, but here's the link:
https://scholarspace.manoa.hawaii.edu/server/api/core/bitstreams/04d7edf5-be1c-4a1e-9c91-995135ac4120/content

[kleene*star]

Can I just ask you what your definition of comprehensible input is? Because if you have to do all that to understand a paragraph (or a page or a chapter), then I wouldn't say it was particularly comprehensible for you.

To be fair, he started this thread to discuss whether reading stuff that's harder than "comprehensible input" was useful, so he wasn't really talking about CI.

Yeah I think I misread it.

[Le Baron]

I think the main thing here is that CI is defined as something you can understand without looking anything up. Anything where you have to look things up isn't CI, and the Krashen line is that this falls into learning about language rather than straight up learning language (he uses the distinction of "learning" vs "acquisition", but that just seems arbitrary semantics to me, and most of the SLA world doesn't make that distinction.)

Whilst it seems fair to state: "CI is something you can understand without looking anything up", the following statement "Anything where you have to look things up isn't CI", is to my mind a reductive false syllogism. This is not really what Krashen said.

I say this because what it boils down to in a real-life scenario is a decision: about whether or not you will make some investigation into the fuzzy edges. It is usually the case that when reading or listening one might not grasp every single thing and this is the case for anything capable of being understood; including information in our own native languages. If you sit through a lecture and then go and read up on what was addressed it wouldn't be because it was 'incomprehensible', but the pursuit of total comprehension. You might choose not to read or look up anything and still 'comprehend' to an effective level what was put forward in the lecture. Some of your understanding involved more than just the 'meaning' and structure of the words delivered.

As you know I am no Krashen superfan (and I think his approach to what output actually does in language acquisition is very misguided), but here is what he actually says about input:

We acquire, in other words, only when we understand language that contains structure that is "a little beyond" where we are now. How is this possible? How can we understand language that contains structures that we have not yet acquired? The answer to this apparent paradox is that we use more than our linguistic competence to help us understand. We also use context, our knowledge of the world, our extra-linguistic information to help us understand language directed at us...

Principles & Practise, 1982

This involves grasping 'meaning' over pure 'structure'. So in essence ALL CI is beyond a person's 'level' in many respects. All reading and listening and communication is a process of meaning and contextual comprehension.

[Tumlare]

The "standard" advice I've heard given by schools and language teachers is "no more than 5 unknown words per page."

FIVE WORDS PER PAGE?

I think the whole thing comes from well-meaning attempts to answer questions about what "Comprehensible Input" really means, because it's something that has been so vaguely defined for so long. 5 is by itself a suspiciously round number, but I've seen criticism that teachers were told to teach "no more than 7 words per lesson", and although I've only seen it as criticism and not actual advice, I imagine it's coming from an overinterpretation of "the psychologist's magic number": 5 plus or minus 2. That was what I remember being told described working memory -- the claim was that the average person can process between 3 and 7 things in working memory and can only identify that many distinct items in view without actually counting.

Regarding the "5 unknown words per page" thing, I think this is probably based on Paul Nation, who says that a reader should have 97%-98% coverage of the words to "read with pleasure", which is quite a different thing from doing an intensive study of a text. Actually, 97% coverage would result in one unknown word every 3.3 lines of text on average from his corpus, so that's going to be a lot more than 5 per page. I haven't read this article in full, but here's the link:
https://scholarspace.manoa.hawaii.edu/server/api/core/bitstreams/04d7edf5-be1c-4a1e-9c91-995135ac4120/content

I don't think it has anything to do with working memory or nice round numbers. It stems from setting a comprehension target of 98% for comprehensible input / extensive reading, as Jeffers said.

The average number of words per page in a standard adult fiction/non-fiction book is 250-300 words. 98% comprehensible = 2% unknown words = 5-6 unknown words per page.

[Khayyam]

On the other hand, if I instead use that hour studying one paragraph (or, page, or chapter--depends on how difficult it is and how advanced I am), reading it multiple times, fussing over the toughest bits like sore teeth, repeatedly listening to a narrator read the passage, and doing my damnedest to be sure I understand every last word (even if a word is so strange that I have to stop and contemplate it for a minute or more to even begin to get the sense of it)--if I do all that, I'll have achieved something concrete, and made undeniable progress.

Can I just ask you what your definition of comprehensible input is? Because if you have to do all that to understand a paragraph (or a page or a chapter), then I wouldn't say it was particularly comprehensible for you.

Oh, no--I didn't mean to imply that a passage qualifies as CI just because I'll be capable of understanding it after I've spent an hour wrestling with it like an alligator. I'm only saying I think it's better to alligator-wrestle than to spend the time reading or listening to something I don't understand and crossing my fingers that I absorb some of it. When I venture into beyond-CI territory, I don't half-ass it--I go in determined to take my trophy, whether that means a sentence or a page or a chapter. The main point of my OP was to ask whether there's any reason to believe I would ever be better served to just to let the blahblahblah wash over me than to attack small bits of it like I do and force it to become CI via repeated readings/listenings. Based on the responses I've gotten so far, I still think the answer is no.

Also, personally I don't think the number of unknown words is that big of a deal if you have a good online dictionary for that language, I've read texts where I basically had to look up half of the words, sure not all of them stuck in my mind afterwards, but like, who cares? A lot of words that you find in written texts are "big words" that nobody uses IRL anyway.

Well, my main goal with any language that I pick up is to develop receptive skills to rival those of a well-read native, so the high-falootin' literary words are just as important to me as the commonly used ones. That's one reason I so frequently go for material that's above me, and repeat it so many times. I absorb a lot of words quite quickly that way--at least, I absorb them well enough that I likely won't have to look them up next time I read them.

Even if your dictionary, or your translation tool or whatever, isn't so good (Google's not so great with Persian), I think you can get along with it because its main job is to just give you a little boost when it comes to learning the word from context. Seeing a word used in many different contexts is the main thing that teaches you what it really means, not knowing the definition or translation.

IMO, the real challenge is actually the grammar. You can look up an unknown word in the bat of an eye, whereas good luck finding the explanation for a grammatical construction that you've never seen before (sure, Google can help out here too, but it's not as straightforward).

Yeah, I agree with that. The best solution I've found for the grammar problem is having an English translation that I can refer to as needed. I can just glance at the English and say, "Okay, this new structure basically boils down to this," and move on.

[tangleweeds]

The looking-up process sucks and is boring (especially compared to reading) any way you slice it, so I like to do all the looking-up for each chapter in one shot rather than break the reading flow to do it.<snip> I tend to think that the things that make it a somewhat time-consuming pain--having to mark the words, having to type in each new word one at a time--improve my ability to remember the words later.

This. I don’t tend to remember words that I only look up—OTOH each additional interaction with the word makes it that much more likely that I will remember it next time. And I like your idea of separating the lookup process, so thanks, I think I will adopt that!

Plus, it gives me satisfaction to have a physical list that I can wad up and throw away with a grin when I'm sure I don't need it anymore.

<vicarious grin>

[Iversen]

When I mentioned flowers names in an earlier posting it was actually a covert reference to the notion of fuzziness, Sometimes you can live with a lot of it - like when you recognize from the context that XYZ probably is a plant, possibly a flower, but it doesn't matter which one. But in some cases it does matter whether 'horned pansy' refers to a tree or bush or a small perennial (or maybe something totally different like a gardening tool) - just as it sometimes does matter whether 'crimson' is a red or a green colour. The problem is that it is fundamentally impossible to predict how much fuzziness you are prepared to accept in a certain situation. And therefore it is idiotic to say that you have to know 98% off all words in a text to accept it as comprehensible. There is definitely a bottom level for what a reader can and will accept - at some point he/she simply can't make sense of anything and gives up. But there is a grey zone where it depends on your degree of knowledge and awareness and patience whether you see a text as 'sufficiently comprehensible'.

I have done a number of wordcounts to assess my level in different languages, and I found out very quickly that there are words which I definitely know and others which I definitely down't know - but in between there are words which I can guess with some certainty or to a certain degree. And I have vacillated betwen different names for that category because it covers different things - but if I meet one of these words in a concrete context I'll probably be content with a vague guess at the meaning and then maybe look it up if it popped up again and again.

As for seeing the syntactic patterns in a text: I have personally mostly studied Indoeuropean languages, but even within that limitation there are languages where it at first can be difficult to grasp how their sentences are constructed (Latin, Irish). But my experience is that I soon will have learned the most common syntactic patterns (also because I do read grammars), and after that it's vocabulary and idiomatic expressions that slow my learning down. Or in some cases semantico-syntactic patterns that differ so much from those you already know that it becomes a problem to produce correct sentences yourself. But even here there is a grey zone. For instance you may recognize a verb in a sentence without being totally sure about its mode - and then you just make an educated guess and go on.

As for just teaching school pupils four or five new words per day ... well, I didn't see that come. It would be horrible - I would have been bored to death if I had been subjected to that limitation during my school time. The number of words you can learn during one hour has absolutely nothing to do with the number of items you can keep in your working memory at any one time.