Chunking...

[Cainntear]

I pulled this out of bombobuffoon's log post 500 hrs by May as I though it was worth discussing further.

(I may have to wind up just faking language skills which seems to be how everyone else does it here, i.e. learning some pre-scripted framework of sentences, and then hope to get very simple answers).
The goal is to be able to watch TV shows and understand 90%. Films also. My reasoning is that if I can understand TV shows (like reality TV) then I should be good enough to begin to converse.

Learning scripts is actually one of the ways to build real fluency, and TV show comprehension will do barely anything for it. Let me explain.

The human brain would be unable to speak any language fluently if it had to retrieve each word, apply the grammar, and then send it to the mouth. Neuron speeds simply aren't fast enough, at least not for regular conversational speed. So how do we do it? Through the magic of chunking. Chunking means storing several words and grammar as one item. For example, the phrase "I'd like a chicken sandwich please" probably consists of three chunks: "I'd like", "a chicken sandwich" and "please". 3 chunks can be retrieved vastly faster than the 6 words of this sentence, not to mention applying the conditional tense to "like", minding the shortening of "I would" to "I'd" and so on.

Right, so I think there's some context needed here. First up, me and sprachprofi are more or less contemporary -- I'm slightly older than her, but I went back to university part-time and was studying linguistics round about the same time she was. There was something called "The Lexical Approach", proposed by a Michael Lewis, that was a response to the Communicative Approach and was kind of in fashion in the 90s, so a lot of universities would have been talking about it in the 2000s. The Open University definitely were, but not really in a good way, the way I remember it. Lewis was basically following the pattern of trends before him: identify one real feature of language and treat it as the "one true king" of language.

1)
The Input Hypothesis/Comprehensible Input took the fact that people would pick up language that they had never been taught as evidence that teaching didn't work.

2)
The Communicative Approach took that as a starting point, but decided that if the problem with CI was the lack of interaction (which certainly was a problem), then interaction is the key to learning a language.
Of course, the problem with Communicative Approach classrooms is that learners can achieve mutual comprehension by... well... murdering the language. (Personally, when studying language courses with the OU, I often found myself in a tutorial with people with very weak skills, and I was aware that speaking correctly would be an impediment to understanding, and I was forced by circumstance to mispronounce things and calque expressiong and grammar over from English. This was not even a CA classroom, but the idea of practising with classmates was always there -- it was not invented by the CA!

3)
The lexical approach is a fairly short logical step from there -- if the Communicative Approach leads to unnatural, non-native patterns, then the best way to start a CA class is with fixed phrases, because if the learners are parroting native-like phrases, that overcomes the problem that errors are more easily understood than correct language.

But as I say, Lewis was overplaying chunking, and that's not the way sprachprofi's talking about it:

Chunking does not happen when watching TV. It does happen when memorising sentences or when speaking. After the brain has laboriously retrieved words often enough, it will develop these shortcuts (chunks). There is no way to develop these shortcuts without having the brain laboriously retrieve words often enough.

Indeed -- and the first evidence of this is kids' language: kids may incorrectly learn very short phrases as a single word, but when I say "very short phrases", I'm mostly meaning compound nouns. In general, though, if they haven't yet acquired the grammar used to build a phrase, they won't be able to repeat the phrase. That's a bit different from adults, because often an adult can be drilled to repeat an arbitrary sequence of sounds.

Imagine being a new piano player. At the beginning you have to look for each key and then place a finger on it and press it. After only a few hours of practice, you know the location of the keys by heart, so the process of playing a single note speeds up. After some more time, your brain has "chunked" notes together that often appear together, so that you can play transitions of 2, 3, 4 notes pretty fluently. This is what active practice does. By contrast, watching concert pianists or piano classes on Youtube will not help with this at all.

And this is a great analogy, because actually the movement from a C to a G is physical "rule", and the movement of G to A is a physical "rule". The movement of C->G->A is a physical rule formed by combining two physical rules and their relation to the "words" of C, G and A.

But this leads to a question of whether you learn those rules generally by only ever doing a specific version of them, or if you need to engage actively in a variety of uses of them, and I think the latter holds, and I believe sprachprofi believes the same. After all, she even goes on to say:

If your goal is to achieve conversational fluency, 60 hours of active practice will do more than 1000 hours of listening.

...active practice: exactly. But then again...

Just look for the "Add1Challenge" or "Fi3M Challenge" (= Fluent in 3 Months Challenge) videos on Youtube. These are people, many of them monolinguals, who wind up being able to have okayish 15-minute conversations, including in non-European languages, after studying 45-60 minutes per day for 90 days. This Youtube list includes my challenge videos for Hebrew, Russian, Croatian, Japanese, Vietnamese. I usually put in 45-60 hours in 90 days (starting from zero) to achieve those and those hours include zero TV, only vocabulary study, some initial textbook study, and a lot of 1:1 conversation practice with tutors. A lecture in which I explain more of my fast-track approach.

Isn't the concept of "chunks" used by the Fi3M sort of philosophy...? I've seen advice from the guy where that term originates advising memorising phrases like "to tell you the truth..." in order to make conversations easier.

See... I don't really get why you even brought up "chunking", as there's not really a clear connection between that and the advice you gave.

Practising actively isn't specific to chunking -- that's how we practise grammar too, and as you seem to be saying, chunking is built on top of grammar, which means we need the grammar first.

And yet your plan focuses entirely on vocabulary, with no importance given to grammar at all. If chunking is important and chunking is built on grammar, shouldn't there be grammar study in there?

[Le Baron]

And yet your plan focuses entirely on vocabulary, with no importance given to grammar at all. If chunking is important and chunking is built on grammar, shouldn't there be grammar study in there?

Isn't that part of:

some initial textbook study

?

[emk]

There was something called "The Lexical Approach", proposed by a Michael Lewis, that was a response to the Communicative Approach and was kind of in fashion in the 90s, so a lot of universities would have been talking about it in the 2000s. The Open University definitely were, but not really in a good way, the way I remember it. Lewis was basically following the pattern of trends before him: identify one real feature of language and treat it as the "one true king" of language.

This is such a common pattern in education, and I've seen it repeated in multiple subjects:

1. A charismatic and brilliant teacher gets good results. They inspire their students, they think hard about how to make progress, and they have some genuinely good insights. Often this is of the form, "Well, we've been underemphasizing X. But if we lean into it, we see clear improvement!" In 90% of these cases, however, the charismatic and brilliant teacher would throw out their theories in second if they saw something going wrong.
2. A bunch of early adopters are inspired, and try the same thing. For a lot of them, it works well!
3. The loose, organic collation of experimenters becomes an Official Movement. Someone writes a manifesto. It gets taught in universities. With luck, this leads to a slight improvement in common practice.
4. The movement becomes an official bureaucratic goal. Management and consultants hand down word from on high. The method gets applied by Boston-area teachers who cannot themselves read at an 11th-grade level. The method is used with students who hate school, who go home to parents who think that education is for pencil-necked geeks. The movement becomes yet another disasterous, idiotic failure.
5. Somehwere, a charismatic and brilliant teacher says, "We could do so much better!" And so it goes...

This is the eternal cycle.

In the computer industry, we spent last 25 years going through a similiar devolution. Kent Beck was a brilliant programmer, who thought deeply about programming. He threw off good ideas like sparks, constantly looking at each situation, and coming up with clever ways to tackle it. He wrote a book titled Extreme Programming, and he inspired a bunch of people to start playing with these ideas. And lots of people saw their work lives improve. Then a bunch of consultants wrote the Agile Manifesto. And then they started teaching managers Agile was the one true way. Then we had "Scrum Coaches." And eventually the Agile movement became a rigid orthodoxy, and finally a shuffling zombie moaning "Braaaiiiiiinnnns..."

I still have massive respect for Kent Beck. My well-thumbed copy of Extreme Programming changed my career. But I want nothing at all to do with too many of the "Scrum Coaches" I've seen at big companies.

The special thing about this forum (and the reason I kept paying the hosting bills when I wasn't here), is that it's a gathering place for independent learners who hang out at stages (1) and (2). People try stuff. Sometimes it works! But there are many different approaches here. If a poster gets stuck, someone else will chime in and say, "Ah, if you're struggling with those bits of grammer, there's a really amazing workbook for C1 students you should try." Advice here can always be changed if it's not working.

This is why I don't spend much time fighting over stages (3-5) above. That portion of the cycle is always painful for anyone subjected to it, no matter how good the initial ideas.

The lexical approach is a fairly short logical step from there -- if the Communicative Approach leads to unnatural, non-native patterns, then the best way to start a CA class is with fixed phrases, because if the learners are parroting native-like phrases, that overcomes the problem that errors are more easily understood than correct language.

But as I say, Lewis was overplaying chunking, and that's not the way sprachprofi's talking about it:

Chunking does not happen when watching TV. It does happen when memorising sentences or when speaking. After the brain has laboriously retrieved words often enough, it will develop these shortcuts (chunks). There is no way to develop these shortcuts without having the brain laboriously retrieve words often enough.

Yeah, there is clear evidence that no level of listening comprehension will automatically result in the ability to speak. There are tens of thousands of "heritage" learners, who've heard a language every day since birth, and who understand it well. But many of them struggle with even basic speaking tasks. One of my friends is an extreme example: She abandoned her native language abruptly at age 5, because nobody at school spoke it. But her mother kept speaking it to her. Today, she can watch gritty cop dramas in her original native language, and even read her family's Facebook posts. But she swears up and down that she can barely do A1-level speaking tasks.

Now, take someone like that, and throw them in a total immersion situation, and those active skills can appear very rapidly! Having a high level of comprehension and strong intuitions for what "sounds right" is an incredible advantage. But something's missing for most people.

(I do have a hypothesis that it might be possible to upgrade passive "earworms" into "chunks" for speaking, using some kind of active recall and speaking aloud. Sort of like how karaoke singers turn earworms into attempted singing. But, uh, I would need to learn another two languages to B1 before I'd want to be more than cautiously optimistic. Get back to me in 20 years. )

[Cainntear]

And yet your plan focuses entirely on vocabulary, with no importance given to grammar at all. If chunking is important and chunking is built on grammar, shouldn't there be grammar study in there?

Isn't that part of:

some initial textbook study

?

Yes, but that's what she says she usually does, and not part of her advice.

I had trimmed off her concrete plan from my post, which was maybe a mistake:

Concrete plan:
1. Use italki to find Finnish tutors and tell them you want conversation practice only and that they should write down every word you don't know in the textchat. Start every sentence in Finnish, even if it's just "I am". Then use an English word whenever you don't know the Finnish equivalent - continue conversing in Finnish. Meanwhile the teacher will write the Finnish word in the textchat. Try to include it in your conversation.
2. Make Anki cards for all words and set expressions that came up this way. Study Anki for 10-20 minutes every morning.
3. When you don't have time for a session with a tutor, imagine conversations in your head, or write down scripts of how you might talk about things that usually come up (your situation in life, how and why you are learning Finnish, the weather, your family, particular anecdotes you often tell...). Practice telling these scripts, don't aim for 100% memorisation but aim for being able to have a short, fluent monologue about these topics without searching for words. This is what Boris Shekhtman calls Language Islands. (Read his book "How to improve your foreign language immediately" for more tips; he ran successful bootcamps to improve people's fluency in just one weekend.)

No mention of learning grammar as part of the advice.

[emk]

No mention of learning grammar as part of the advice.

In my log, I don't talk a lot about studying grammar. And yeah, my idea of a grammar book is a laminated quick reference sheet or a US$6 Dover Essential Grammar book, which has maybe 50 tiny pages of real content.

But I have a couple of things that help me out with grammar:

1. I am a bit of an outlier. I write parsers at work! (Most recently this past Thursday.) And I've studied grammar in English and Latin. So like almost everyone here, I know how grammar works.
2. I take an interest in little details. "Oh, so we attach a -lo suffix here in Spanish. It's another language with sticky pronouns that glom onto verbs! In French they go on the front. Cool."
3. When I see something interesting, I dump it into an Anki deck. And I review it until it looks 100% natural. I may not know the rule, but I'll be able recite the example phrase in my sleep.

So even though I'm completely half-assing grammar, I still get a lot of grammar done in some form. And I focus hard on activities that build my sense of what "sounds right."

And this is exactly the sort of thing that gets "lost in translation" when giving advice. It's certainly possible to learn a grammar with a mix of 70% osmosis, 25% noticing curiosities, and 5% looking things up. But other people may get great mileage out of sitting down with a good grammar book and working through the exercises. One of the most valuable skills a good teacher brings is the ability to diagnose what's going on with a student, and to make focused suggestions.

Having a single, inflexible theory is ultimitately limiting. And it makes for worse teachers and mentors.

[MorkTheFiddle]

Fascinating thread. Reminds me of some fruitful discussions in HTLAL. Let's have more.

[Axon]

Has anyone done research on how accurate the listening of self-described receptive bilinguals is? Sfuqua wrote something a while ago about his passive ability to understand Tagalog, where he can watch TV and understand it fairly easily. During a short period of intensive Tagalog study, he paid close attention to the details of the language in the TV shows he was watching and realized that he was actually missing more than he realized. Something similar happens when I read a book in my native English with a lot of archaic or rare words. I understand what's happening, but the more unknown words there are in the description of something, the more likely it is that I've missed something relatively important to what the author was trying to convey.

I think my point is that I'm wary about using heritage speakers as evidence for the claim that people can achieve extremely high passive skills with nearly no active skills. My hunch is that, like me reading Moby Dick and skipping all the obscure sailing terminology, someone watching TV in a heritage language with no speaking ability is likely missing more of the language than they think. I don't want to make any counter-claims of my own, but I'd like to see this one investigated more thoroughly.

[bombobuffoon]

So is "chunking" just adoption of phrases? If so why call phrases "chunking"? Or have I missed something.
It seems "chunking" is some kind of term in psychology as well. Maybe they share the same origin.

Anyway some words are used very often together to mean one thing and I even think of it as one word. I don't know that its really worth over analysing it but it just seems to be plainly obvious. And this is just something I pick up over time particularly from CI.

For example toiselle puolelle, on the other side. Even google translate has this marked as a verified phrase so this is a commonly recurring pattern.



I don't need figure out the conjugation of other and side, I just have that phrase in my head for situations when I need to describe doing x on the other side I have that. Of course there's times that won't work at all but its a useful combo.

Going from words to patterns or phrases means I am getting more used to thinking in the other language rather than English. Just small steps of course but helps me "think" in the other language. On the other hand if trying to invent language and just explicitly stating everything even if grammatically correct the result is gonna be pretty damn weird. There needs to be a body of knowledge from which to draw phrases from.

Indeed the textbook I use takes a very practical approach to adopting grammar, and is always pointing out these kinds of common patterns such as the above but for me what sticks the most are the ones I notice myself with CI.

[bombobuffoon]

Has anyone done research on how accurate the listening of self-described receptive bilinguals is? Sfuqua wrote something a while ago about his passive ability to understand Tagalog, where he can watch TV and understand it fairly easily. During a short period of intensive Tagalog study, he paid close attention to the details of the language in the TV shows he was watching and realized that he was actually missing more than he realized. Something similar happens when I read a book in my native English with a lot of archaic or rare words. I understand what's happening, but the more unknown words there are in the description of something, the more likely it is that I've missed something relatively important to what the author was trying to convey.

I think my point is that I'm wary about using heritage speakers as evidence for the claim that people can achieve extremely high passive skills with nearly no active skills. My hunch is that, like me reading Moby Dick and skipping all the obscure sailing terminology, someone watching TV in a heritage language with no speaking ability is likely missing more of the language than they think. I don't want to make any counter-claims of my own, but I'd like to see this one investigated more thoroughly.

This is very interesting. I have had exactly the same thoughts.

Kind of related also, there are periods when I "feel" I am understanding far more than I actually am. It can be a queer effect to experience. It happens to me after long periods of listening and usually with tiredness. I will be listening for maybe 20 minutes thinking I am understanding and then snap realize...wait what's going on here?

I suspect a lot of people who "understand but don't speak" may have this similar effect.

[Cainntear]

So is "chunking" just adoption of phrases? If so why call phrases "chunking"? Or have I missed something.
It seems "chunking" is some kind of term in psychology as well. Maybe they share the same origin.

Yes, absolutely.

There was discussion about the difference between the "open choice principle" and the "closed choice principle" relating to Lewis's lexical approach.

The closed choice principle is that only certain things are permitted -- a finite list of candidate words for a slot.

The open choice principle is that there are slots were almost anything is allowed.
Consider: I shot myself in the foot, you shot yourself in the foot, they shot themselves in the foot, the guy on the train shot himself in the foot.
__1__ shot __2__ in the foot
That first slot is an open choice -- the second is an extremely closed one.
But consider also that in they shot themselves in the foot, I couldn't imagine anyone ever saying "*they shot themselves in the feet" or "*Each of them shot himself or herself in the foot... there is no choice at all about the foot: "in the foot" is a fixed phrase within a (semi-)fixed phrase. Lewis's lexical approach was about treating phrases as being basic vocabulary units -- essentially thinking of "inthefoot" as a word.

This is where the psychological term chunking comes in -- the idea being that there is a single overarching concept that can represent the whole thing in working memory to save space and activate the subordinate concepts it is made of.

This was an idea that was thought of more in language after Lewis, with chunking addressing the good and bad of the lexical approach.
I've seen chunking treated as though it supports Lewis, but it doesn't -- not everything is a word. But there were definitely people wedded to the lexical approach that saw it that way, and "chunking" is often abused as a result.

I have no reason to believe that sprachprofi falls into that camp, but she has fallen into the trap of using a technical term without explanation, and that doesn't help clarify anything.