Just FYI, it still seems like I need to login at least once per day on both my phone and desktop browsers.
zKing's recent posts
leosmith wrote:I ask because I've done some informal documenting of time I've spent using certain techniques in my learning, and I've found that my improvement in conversation is almost directly proportional to the number of one hour conversation where I briefly noted and later memorized vocabulary and sentences that I lacked or didn't understand. Like yours, my classes were also 100% target, and other than the noting process which only took a few minutes, were just as you described; no corrections unless they truly didn't understand me.
To be perfectly honest, I've been using this method for a long time now, and the only things I compared were periods when I read 2 hrs/day vs 30 min/day, periods where I did the exercise mentioned above vs conversation with no memorization and periods of 5 - 7 hours conversation per week vs 3 or less hours per week. As I mentioned, the overwhelmingly best indicator of level in conversation turned out to be the number of hours spent conversing and memorizing. I was quite surprised that 2 hrs of reading wasn't noticeably better for conversation than 30 minutes for the single data point of myself.
Wow, that is REALLY interesting to me... and makes sense. You've just sparked my motivation: I've been taking a break from Anki for a couple of months now, but I'm going to restart and only use production cards from items collected from speaking gaps in my iTalki sessions (or other production attempts). I've had the thought in the past to narrow my flashcards to only items for which I actually experienced a missing need in production... but this nicely put the cherry on top of that idea.
You have way more successful experience than I do and I really appreciate you sharing this.
I didn't have a consistent repo with this bug, it was somewhat rare and intermittent, but I also haven't seen it in a while so I would consider it closed.
I'll ping this thread again if I ever see it again.
Thanks for all the hard work!
Confirmed! Everything is working great.
Confirmed! Everything seems to be working great. I added this text (containing all known issues) and it worked fine.
00:00:04,835 --> 00:00:07,657
大家好， 歡迎收睇垃圾驚批鬥座， 我係瘋子愷
Great! I'll have a chance a little later today to test myself and I'll let you know. Thanks!
See the other thread, I'm getting the spinning Save button and all Chinese characters are converted into ? marks.
Even in the forums. Here is some Chinese I'm typing into this post's edit box: ???????
I guess the only way to tell will be after a couple days (after one more sign in to get another cookie, I'd assume).
I'll let you know!
I just checked with another tiny Cantonese passage (even with no SRT tags), yep spinning Save button and all Chinese characters are getting turned into ? marks.
So it looks like no one can add Cantonese passages right now.
Sorry, but it looks more broken than before.
I created a test passage with the following text:
00:00:00,000 --> 00:00:03,260
(Edit: Whoops! It looks like I can't post Chinese here either? They are all turning into question marks.)
When I saved, I got the forever spinning Save button.
When I tried to edit the new text I saw:
00:00:00,000 --> 00:00:03,260
It looks like a regression.
The passage name is "Test Numbers With Commas".
I've attached a small SRT file that would be a good to test with.
Thanks again for the kind comments.
leosmith wrote:I agree that we need comprehensible input, and tons of it, to learn a language to a decent level. Without it, we fail. But when he says nothing else facilitates acquisition, I disagree. Sure, there are some people who never study grammar, vocab, etc and reach high levels in a language, but I think they are pretty rare.
Yep, I agree with you. But I look at those other things as amplifiers of input and what makes input comprehensible. But it certainly isn't a distinct line. For example, when someone "learns" a word like 'bianco' in Italian via flashcards and later they encounter the word... they think for a second and say "oh, I know that one, it means 'white'". I can't argue that they haven't 'learned some Italian' in some sense. But when they claim they now 'know' that word and 1000 more in that same way, but they can't understand most sentences that use only words from that set of 1000... there is this disconnect. If they continue down that path, and collect another 1k, 2k, 5k words... they start to miss the point. If you can't understand and produce Italian to communicate or read or listen to the radio, etc... at some point you are a human dictionary, not a speaker of the language. Of course I'm exaggerating to make a point, but it isn't that uncommon to find folks who say they are learning a language and the ONLY thing they are doing is flashcards or some app that is pretty much the same thing.
Flashcards are great, I use them. And you are right, I'd bet that MOST successful language learners do as well. I just advocate, at any opportunity I'm given, that people recognize that they are a means, not an end.
And I totally think Krashen's single-mindedness on Comprehensible Input misses the boat on output, particularly with speaking. You can get all the input in the world, but at some point you have to practice speaking or writing if you want to have those skills. They are just a different animal than reading/listening. Pronunciation and prosody are skills of their own. Mechanically making your mouth do the right thing can be hard. You can 'kind of understand' your way through a lot of input and do just fine when listening or reading... but when speaking you are all on your own to make every little word.
Krashen has created a very good hammer for learning input with his "Comprehensible Input" theory, but unfortunately he sees speaking and writing as just another nail and I don't think he's right.
leosmith wrote:Just curious - do you do conversation lessons with tutors, take notes, then put them in anki?
For Cantonese, I hadn't attempted any real output for nearly ten years (my giant mistake #2)... mostly because of my Krashen-like over reliance on input. A couple years ago I had an enjoyable detour with Italian, where I finally learned how important output practice was (among other things) and did weekly iTalki sessions. In May of 2018, when I decide to pick up Cantonese again, I started regular iTalki sessions and a bit of journaling. I take notes, but I honestly don't do much with them. I don't put anything in Anki from them. Up until recently my iTalki sessions were 100% focused on speaking fluidity. I told my tutors to only briefly correct me if I was unintelligible... let me mangle things, but just let me practice speaking as much as possible. I would do at least 80% of the speaking for most of the sessions. I went from stammering and sputtering a few words here and there, with lots of long pauses, to being able to mostly hold a conversation... don't get me wrong, I still murder the language and make a constant stream of mistakes. But I can communicate, be understood and have a semi-interesting conversation for 30-60 minutes. With my current tutor I'm now having him focus a lot more on my accuracy.
Thanks, leosmith, for the kind words. I think we have very similar learning styles in many ways.
leosmith wrote:wrote:(ask me how I know)...
How do you know how?
Hehe, I guess I left that bait there, didn't I?
I have a bad habit of 'preparing to learn'.
For hours on end I would OCR text from books, correct formatting of text files, rip DVDs, process Mp3 to strip dialogs of introductory music or titles, read blog posts of polyglots that say something I've already heard 20 times before, develop memory palaces for vocabulary, search the internet endlessly for 'better' content, carefully curate Anki decks with TTS audio and closures, write software to 'speed up' all of the above tasks, etc. etc. etc.... and still fail to take 15 minutes to read/listen to that bit of content I already had ready to go.
At some point not so long ago, I was hand writing out some Chinese characters and it hit me:
My real goal is to have fluid conversations in Cantonese. Hand writing characters, while fun and interesting, was VERY far removed from actually accomplishing my goal and ate a lot of what little time I had to study languages. And many of the other things I was doing were more or less in the same boat.
I've curbed this habit significantly, but it is still something I am careful to watch out for.
I strongly believe in Krashen's Comprehensive Input model (for input at least).
And to me, for Listening/Reading: Language Skill = Input Comprehension Level * Total Input Volume
Or put another way: If you want to understand speech or text in your target language, you must attempt to listen or read and ensure that input is at least somewhat comprehensible.
And to learn to Speak or Write, you must attempt to Speak or Write... a lot.
Everything else in language learning is just scaffolding for the above.
Vocabulary flashcard decks, memorized grammar, etc are all great ... if they are directly increasing your comprehension of the input you are currently getting or allowing the output you are currently creating to be better.
But if they are words/grammar you'll need 'someday'... or worse, building these tools is eating a lot of time you needed to actually read/listen/speak/write....
Then building these tools may not be needed right now (or perhaps ever?). Note that, ironically, many of these bits will likely be FAR easier to learn, or even come collaterally, when you have built a solid base of skill in using the language.
It is very easy to end up like that guy who has a workshop stuffed to the rafters with tools, but he's never actually built or fixed anything.
He is constantly collecting tools and organizing his workshop, but never seems to find the time to actually do the task he's been preparing for.
It is very easy to be doing "something" related to your TL and think that this is progress... IMO, if that "something" isn't reading/listening/speaking/writing at least full sentences and more in the TL, it may be generating a lot less progress than you think. (Note that I DO believe in an initial bootstrap period where you learn some basics of pronunciation, grammar, and a set of the most common words.... I'm talking here about after that beginner phase.)
I've been learning Cantonese off and on (more off than on) for more than 10 years now and I've probably taken every wrong turn there is along the way.
I consider the above issue to be the worst of them.
I have made FAR more progress on my goal of being conversational in Cantonese after I got really focused on actual (intensive) listening and speaking as my absolute number one tasks and I put all that other stuff in their proper secondary roles.
That said, after all that 'sounding like an expert' talk, I should caveat with:
I'm now a solid B1 in Cantonese and a rusty B1 in Italian... so I'm far from the most experienced language learner around.
It seems like I need to re-sign in to this site daily on both my phone and desktop browser.
Most other sites of this type seem to expire their sign in cookie in a week to a month.
It is somewhat annoying, particularly on the phone browser.
Thanks for listening!
Yes, I've played with the memory palace technique (and many similar) in the past. It does work, however, as mentioned above it is incredibly time consuming.
I don't recommend it for language learning as you tend to lose sight of the forest for the trees.
IMO memorizing a zillion short definitions of words in your target language isn't the goal. This is merely a tool that helps you along the way. The real goal (for most) is to be able to fluently use the language, both understanding it and producing it. Successful language learning isn't merely memorizing more and more grammar rules and vocabulary items until someday you have 'enough' and you are suddenly fluent. Memorizing this stuff gives you a rough grammar cheat sheet and a (very short, poorly edited) dictionary in your head, sort of like a pop-up dictionary on your computer. It is the act of USING these clunky tools repeatedly, over and over on real language, decoding the language and generating bits of it with lots of conscious effort, that eventually builds that automaticity/fluency in the automatic part of your brain that we are really after.
Note that many successful language learners will use tools like the Reader tool on this site and don't even attempt to put special effort into memorizing grammar and vocabulary. I'm not saying that memorizing that stuff isn't efficient if done with some restraint, but just that it isn't completely necessary on its own and it isn't really the end goal.
It is very easy to get overly focused on the acquiring or creating these kinds of tools and forget to do the real work to actually learn to use the language (ask me how I know)...
At a high level, for someone who hasn't learned a language before I'd recommend month long 'sprints' or cycles:
Try a routine for a month/4 weeks and see how it works for you. Each month, keep what works and change what doesn't.
A month is long enough to experience your process, feel its effectiveness (or lack thereof) and see how it fits into your life, but a month is not so long that you over-commit yourself to a bad idea... if you are someone who tends to have a problem with that.
Personally, if I were to start a new language, I'd spend roughly the first month or two doing a 'boostrap' to get a foot hold on the language:
1. Read a good explanation of the pronunciation (familiarize but don't attempt to memorize it)
2. Read/Skim a reasonable description of the grammar (familiarize but don't attempt to memorize it)
3. Anki/SRS the first ~500-1000 most common words
4. Start listening to the language in my dead time (e.g. while driving) with no expectation of understanding much, just to get used to the sounds.
5. Work my way somewhat casually/rapidly through a beginner text book like Assimil or Teach Yourself. Mostly I'd just listen/read the dialogs and decoding them and notice the nuts and bolts of how the language functions in practice.
After that first 1-2 months, I'd jump in headfirst, I'd...:
1. Try to find as much (hopefully interesting) Audio+Matching Text as I can.
2. Use the Reader app here, as much as I can, intensively read the above content (and anything else I find interesting)
3. Use a free app like WorkAudioBook or Audacity to listen to the above content while reading in the Reader. These audio apps allow you to easily short-loop phrases/sentences repeatedly so you can listen to each phrase until you _really_ hear it.
4. Get a tutor and try to start speaking and having conversations... and secondarily use them to answer questions that arise in my reading.
5. (Edit) Oh... and depending on your goals, start writing short journal entries in your TL and get corrections on them.
Some like to continue to use Anki/SRS and attempt to memorize simple definitions of 1000's of words, but I've found it doesn't work well for me. For me, intensive reading of interesting Comprehensible Input solves the same problem (and simultaneously many others) in a much more enjoyable way.
Lots of luck to you all!
For me: Making time. But secondarily I would say it greatly depends on the language.
Cantonese: 1) The Spoken Language and the Written Language are very different 2) Finding Content 3) Tones (both hearing them and producing them)
Italian: 1) maybe the Grammar? ... for me, after learning Cantonese for a while, Italian doesn't feel like it has many obstacles. It is just a matter of time on task. :-)
Thanks! Yeah, I could strip the SRT timing stuff, but I'd like to leave it in so that, if I share the text here, others would be able to use the timing either manually to see where they are in the video, or they could just copy/paste the file out and use it in other tools like WorkAudioBook (my fav for loop-listening to content)
Text like the following is being treated as a 'word': 28,543
The work around is to "ignore" them, but this is annoying as I'm importing SRT subtitle files that have lines like below, so I'm getting a LOT of them.
Thanks for listening!
00:00:12,473 --> 00:00:15,409
I just noted that the issue with spaces seems to be fixed, thanks guys!
Odd Chinese characters is still a problem, but this is a big step forward in easy-of-use.
(I know that the team is already aware of this issue, but I wanted to post this here for tracking and so other Cantonese learners are aware of the problem and the temporary work around.)
If you attempt to save a Cantonese passage with spaces in it or some rare Chinese characters, the "Save" button will spin indefinitely and the passage won't save. The rare Chinese characters also seem to cause failures to save when used in custom definitions and forum PM's.
The temporary work around is replace these characters with others (i.e. replace all spaces with a dash "-" for example).
I was working with a longer Cantonese passage and after doing numerous joins, at some point right after doing a join, the text became slightly mangled such that some of the text was duplicated and the highlighting was over two words/phrases at the same time (see attached pic and original text below).
If I click around, it would highlight two words/phrases where ever I click.
If I exit the passage, the stats then show the passage as having 0 words. If I go into the passage it will have no text and the stats show 0 words.
If I then EDIT the passage, the original text is in the edit form in its original state.
If I trivially alter the text in the edit form (add a new line) and Save, the passage will be "fixed", but I lose all of my joins/splits and have to start over.
This has happened a couple of times in the same passage, but it doesn't happen in the same place in the text. Sometimes it happens quickly, other times it takes quite a while to repro.
To hopefully help with finding/fixing the bug, I've now made a copy of the passage and repro'd the bug and left that passage in the broken state.
It is named "Test Long Passage".
The screenshot is right after I joined the highlighted word 兼且 and the bug appeared, you can see the duplication there and for the lines below it.
Thanks for listening!
Here is the original text to compare with the screen shot: