Implement "continually process information at speed (Imron)"

[querido]

Let's talk about setting up the tools and a workflow to drill to "continually process information at speed over a sustained period (Imron)". This is not restricted to listening comprehension but I hope to focus on that in this thread, as reading at length is already very well supported by e.g. Chinese Text Analyser. In this post I will quote Imron's rationale, and then we can talk about how it could look in practice.

 

Imron spake unto us, saying:

 

"If you practice drilling flashcards then you'll get good at drilling flashcards...

...to get good at reading you should do reading

...to get good at listening ... you should (listen)..." 

 

and:

 

"Flashcards can help as a revision aid, but they shouldn't drive the learning process. There're too many things they don't drill you on - especially with regards to ability to continually process information at speed over a sustained period..."

 

The question then, is, how should I apply this principle in my own studies; what would the tools and workflow look like? I'll share some ideas, as I tried this in the past, but it became the iceberg to my Titanic, unable to turn. But so much time has passed that I've probably escaped addiction to my old flashcard decks. Haven't touched them. Lately. (Holds up hand) I swear.

 

What should the tools and workflow look like? Thank you. :-)

 

[querido]

#1 (manual method requiring no special software)

 

Number your whole-lesson audio and text files with a due date and interval (only if desired, so they stay in order in a file manager, like this:

151223-1-Lesson X.wav

151223-5-Lesson Y.wav

151224-3-Lesson Z.wav

[Edited: No need to rename the text files. The text could be on the "back" to be viewed if necessary, or all texts could be collected in something that supports studying them, like CTA, and just search on the name of the lesson.]

Then just click on them and study. Rename the files as you think necessary when you're through with each. That's a little tedious and troublesome. It *is* no-brainer simple, though (and at this moment it still looks attractive to me!). I said "whole-lesson" so you have maybe 10-100 of these in play, not thousands.

 

Refinement: the audio and text could be combined in e.g. an .mp4 container and played in a player that will show the text. I had this working in foobar at one time. But, better might be if the text is displayed somewhere where the words could be looked up as necessary. Can a web browser be set as the player for .mp4s and work like this? But, better would be if the text is displayed in e.g. Chinese Text Analyser which can keep track of which words you've looked up (so you don't have to stop listening and fuss with text). I don't want to talk about flashcarding them now, but the wordlist produced could be reviewed some more the same day, or not, after all of your listening is done.

 

Very important and convenient note: The above is not the faux-rocket-science where you aren't allowed to listen to the next lesson before it's due. So the whole directory of audio files could be set on random/repeat and allowed to play as continually as desired, when you don't have time to attend to it.

[querido]

#2 (more automated)

 

Imagine Anki and Chinese Text Analyser set up side by side, or imagine that CTA has aquired a flashcard brain and audio file handling. 

 

Need not call it a flashcard, but the "front" would play a whole-lesson audio file. Look at the text only if necessary. Look up a word only if necessary. Listen as many times as desired and score the card however desired.

 

Anki can display text, but I said CTA because it can make a wordlist for you without interrupting the listening.

 

More notes: The flashcard functionality would merely do the scheduling automatically. You need not be bound to those numbers and you would still have a directory somewhere with the audio files which could be played on random/repeat, if desired, when you aren't attending.

These lessons, which could be paragraphs from a book for example, could be collapsed someday into whole-book lessons. If they are unbearably easy by then just set the interval to a year or whatever and press on.

[querido]

This space is reserved for future strokes of genius. Edited:

 

#3 Fully automated

 

Someone - with sufficient power of inducement - should commission Imron to add this to CTA or to write a separate, somehow audio-mode analogue. We should start a wish list of features. But it should facilitate doing as he advised, which I quoted in the first post: "continually process (audio and text) information at speed over a sustained period", while avoiding tedious entanglement in the usual flashcarding-thing.

 

I hope that others will share their ideas now. Thanks a lot! :-)

[Shelley]

It seems to me that you are only swapping your old flashcards for a new "disguised" set of flashcards.

 

I would say to improve reading - find texts you think you might enjoy reading, put it in CTA and if it is just a small amount above your comfort zone then read that referring as little as possible to a dictionary.

 

Put any new words you come across in to a flashcard program, Pleco for instance, and study the new words. when you feel you "know" them regardless of what the flashcards say, stop with the flashcards.

 

Then repeat step one, find a new text and continue the same way.

 

Create a a new deck of flashcards for each text, and only revise them for a while.

 

After a few texts, depending on the length of the texts, go back over the flashcards and make a new deck with the cards that just don't stick, but also apply a bit of "weeding" if there are specialized words that you probably will never need, don't stress learning them, you may if you come across them again later for any reason say hey I remember this one or not, you can always look it up, if its frequency changes, then put it back on the list.

 

Follow the same principal for audio.

 

Use flashcards wisely and sparingly.

 

Read and listen lots then read and listen some more

[Tianjin42]

Relating to the previous point, what sort of 'percentage known' do you guys aim for on Chinese Text Analyser?

As in - an ideal percentage; or perhaps you think of it as an easy, medium and difficult setting?

[querido]

Thanks Shelley.

I already know how to study reading and writing, where texts get scheduled for review.

But when working on listening comprehension, it is audio files and the text corresponding to them that would be scheduled for review.

This scheduling could be done simply and manually, as in #1 above.

Or, Anki could do it as in #2. View it not as flashcarding but as just automatic review scheduling. 

 

Edit: "...I hope to focus on (listening comprehension) in this thread, as reading at length is already very well supported by e.g. Chinese Text Analyser" and by threads already devoted to CTA and how to use it.

[realmayo]

Tianjin42:

 

For percentges, see #296 here: http://www.chinese-forums.com/index.php?/topic/44383-introducing-chinese-text-analyser/page-15

[realmayo]

querido: I have a system which provides audio comprising new material, material I've heard a few times only, and material I've heard plenty of times before.

It took a while to set up, fiddling with playlists. But now it's practically zero maintenance. When it's time to listen to something I just press play. I don't need to waste time worrying about finding something suitable or wondering whether and when I should relisten to a certain track.

It's kind of SRS for audio tracks.

Here: http://www.chinese-forums.com/index.php?/topic/47516-smart-playlists-for-varied-repeated-listening

 

 

Before, I was faffing around too much working out what to listen to, or what to relisten to. Silly as it sounds, now that I can let the PC make all the decisions I’m actually I’m listening to far far more: I’ve got Chinese on all the time, either listening to it hard, or having it on while I cook or tidy up or whatever.

 

Edit: I'd assumed you could just read the associated text when the audio came up, but re-reading your post it looks like you're after something more integrated.

 

The above process simply means that you're guaranteed to re-hear audio regularly into the future. So if you've worked hard on the track the first time you listen to it, hopefully much of what you learned then will be recallable once you relisten, as long as that's not too far into the future.

 

I wonder if you couldn't be just as effective by spending a decent amount of time on an audio file when you first listen to it; excerpt key vocab and tricky phrases; and have them ready to refer to

[querido]

This could be done manually but one can imagine being walked through it automatically:

 

Imagine a full-length book, like a novel, at a suitable difficulty level, with audio.

Imagine a window into this book - a segment of the audio/text containing some manageable number of unknown words. (Manageable might mean too many to learn all at once, but set so that you can be exposed to some as you're learning some, maybe 10?-100? new words.)

Now, the audio of this section repeats mercilessly as you study the words and try to understand all of this audio "at speed" (meaning in real time).

Now, the window will move. The endpoint will stretch farther into the book and the startpoint can begin later, as you've proven - by satisfying some set criteria - that you understand everything that came earlier.

 

Refinement:

Instead of looking at a time scale (which means nothing) in the music player and moving the start/end points, or instead of clipping out these sections of audio, imagine that your text editor/reader (which is supporing the learning of words with integrated dictionary and wordlists etc, e.g. CTA) could reach into the audio. For example, looking at the text, right-click/Start_Audio_Segment_Here, right-click/End_Audio_Segment_Here. (Hey, it would be nice to have stats for unknown words between those two markers.) Finally, for extra credit, imagine that it could move those start/end points automatically (iaw some chosen criteria).

[querido]

To realmayo: I remember that post! I'll take a look at it! :-) [Edit: I did set up a system based on your post in the past. I just started iTunes for the first time in months and a system based on yours is, conveniently, already in there. :-) ]

 

It doesn't matter if it's coarse-grained like that. As I said, we need not shoot for full faux-rocket-science. :-)

 

In addition to the above I'm also interested in looking inside an audio file, coordinated with e.g. CTA, as I tried to talk about in the previous post.

[realmayo]

This would actually be great: less, for me, as a way to integrate reading and listening skills -- because in practice, these two skills aren't usually combined (well, I guess they are on Chinese TV) -- but more to speed up the process of looking up difficult passages in a transcript, which can get extremely frustrating.

[Shelley]

@querido I didn't mean to you tell how to suck eggs just sharing one way I try to do this.

 

I hope you find a way to help schedule reviewing.

[imron]

I have a bunch of comments to add, but am busy with Christmassy things for the next few days and won't have to add them until then.

In the meantime, checkout things like the chairman's bao, and du chinese which both provide synced text and audio aimed at a variety of levels.

[querido]

To Shelley: Yes, I understood that! Thank you. :-)

*

I didn't have in mind integrating reading and listening skills (realmayo), nor necessarily synced text and audio (Imron), but I'll check those sites, thank you.

[querido]

In the original post we said  "flashcarding... [of words but probably sentences too] ...should not drive the learning process (Imron)". So I immediately think: How do I make listening comprehension "drive the learning process"?

 

Maybe my first thought is to put the entire lesson audio on the front of a flashcard and open the text in e.g. CTA and study only what I need to, promoting or demoting that lesson accordingly. This should not be seen as theroretically-orthodox flashcarding, but just as automatic scheduling. That's all.

It is not necessary, but suppose Anki could autorepeat that audio. Then there you would have, most simply, intelligently scheduled "continual processing of information at speed (meaning, in real time) over a sustained period (paraphrasing Imron from the original post)", for as long as you can stand it.

[Edit: But material that is inconveniently long like a graded reader must be chopped up I guess. THAT practical necessity is part of the reason for some of the other posts, above, which have a more generalized solution in mind, I guess. Generalized would mean that the entire corpus/repertoire/collection would be in CTA (text) and Anki (audio), with some kind of coordination between them. As I said, the coordination could simply be manual. Anki could be seen as serving as an audio player with intelligent scheduling.]

 

I already thought this many times but never actually made it work for me; I always drifted back toward orthodox flashcarding, whether words or sentences. Thus, the thread. Like realmayo, you might have ideas about it.

[querido]

In the previous post I said that CTA (text) and Anki (audio) would be coordinated. Here's one way that could be done simply enough: if the audio isn't enough and you must look at text, then answer the card and see the lesson text on the back. If that still isn't enough then Select/Copy a chunk of that text and search CTA for that to get to it in CTA. 

 

Now I need Anki to autorepeat because I know what will happen if it doesn't; I would go over to CTA for a while and not want to keep pressing the play button. 

 

A session with Anki as described in this post and the previous one would look like continuous listening comprehension exercise/drilling, like "continual processing of information at speed over a sustained period", as required by the original post, only looking at and looking up text as needed, which is unavoidable.

 

I want to point out a dreamy possiblity. CTA can keep track of the words you've looked up and can make lists etc, which usually lead to orthodox flashcarding. But wouldn't it be really neat if one could, in conjunction with the above, renounce orthodox flashcarding completely? I'd avoid getting tangled up in that stuff again. OR, one could make only writing cards (to remember how to write).

[querido]

If you want autorepeat and other audio handling functions provided by your favorite player, then it could look like this:

 

1. Anki with the name of the lesson on the "front", to tell you what to study and when.

2. Favorite audio player with whole-lesson audio in a playlist.

3. CTA with the text of all the lessons.

 

Lesson names should be preceeded by a number to keep them easy to find.

 

Anki + Foobar2000 + CTA = listening comprehension driven study program to "continually process information at speed over a sustained period", as prescribed. And whenever you aren't studying realmayo's setup could be playing continuously. Am I done here? What do you think?

[Tianjin42]

@realmayo

Thanks for the link.

[querido]

Listening comprehension "driven" must mean:

 

1) The desire to understand *this* sound flying by drives the learning of *these* words (not some list of words I'll need in the future).

2) Learning new words is driven by the need to understand *this* stream of sound (not one I expect to hear in the future but the one playing now as, logically, I cannot be driven by something I haven't heard yet).

 

So, assuming we're reviewing enough, a mountain of poorly-known word cards could never accumulate.

If you say "yes, but a mountain of poorly-understood audio files would accumulate" I would reply that at least the latter is inarguably real language whereas the former is still distantly removed from useable language (which I know from personal experience).

 

Some would argue that sentences are a happy medium. I have an answer to that. To "continually process information at speed over a sustained period (Imron)" with sentences as the materials we would be listening to a string of disconnected sentences. That sounds unpleasant and doesn't include all of the connective material in dialogue, narrative, etc. Can't be ideal listening material.

 

Single-word, "minimum information", etc, has a high priest. Sentences have at least one guru. ONE reason why a listening comprehension driven program with more extended materials is so infrequently discussed is that the "grasshoppers" don't have a "master". I don't aspire to that either.