This topic contains 21 replies, has 11 voices, and was last updated by 
-
Posts
-
Did you grave-dig this thread just to reminisc? =P
Congrats for getting the top spot on Google, though. =)
Partly. And partly because other, newer members might find it interesting, like I said :P Would anyone actually find this interesting, I wonder…
I’m not a grave digger, I’m a necromancer.
I’ll likely have a look at these at some point.
-Cheers!
> Wonder how Sheepy, Mark, Armando, jkl, and winterpromise31 are getting on these days.
I’m still studying every day, and I still believe in the sentence method. Having spent more time with it I would say that the order in which you study the sentences is vitally important, and a key part of the method. You really need to find some way of sorting them so that you don’t introduce too much new stuff at once, especially when it comes to introducing new kanji. In particular, the sentences in the Core collection aren’t ordered very well for study using the sentence method.
I tried a few different approaches, end ended up ordering the sentences so that sentences which use words with the most common kanji readings appear first. To do that you need to know the reading of each kanji in each sentence, and that is going to require either some programming ability or a whole lot of manual effort. A simpler approach that doesn’t involve programming would be to use something like the below Anki plugin, which is designed to reorder your deck based on word frequencies.
http://rtkwiki.koohii.com/wiki/Morph_Man
I haven’t used it, so I can’t really say much about it. But if you do decide to try it, you should definitely make a backup of your deck first just in case something goes wrong.
Edit: Sorry for the essay, oops >.<
Nice to hear from you jkl :)
I think ordering only really matters if you’re using a big set of pre-made sentence cards (like with Core 6k). If you’re building a deck from scratch – as I intend to do – you have total control over the difficulty of the sentences since you are the one picking them i.e. you specifically pick ones that are just the right amount of difficult, not too easy, not too hard.
There are a few other benefits to making a deck yourself, at least from what I gather from reading through chunks of AJATT: you can choose sentences that are interesting or from material that interests you (knowing the context of the sentence can help with memory); sentences don’t have to be full sentences like in Core 6k, they can be expressions or short phrases or clauses; the process of manually adding the sentence makes you remember it more easily; you need to actually read/listen to native material to get the sentences, which is beneficial in itself; you can format the cards in any way you want, including having J->J cards.
Back around the time I first made this thread, I actually tried using a sentence deck made from Core 6k. I found that a lot of the sentences for words that I already knew were boring and I ended up trying to delete so many – I knew it was going to take a while to do, so I eventually lost motivation and gave up on the idea of sentences completely for the time being. When you have tons of sentences like「AはBです」or「Aが好きです」, there’s really no point in reviewing them XD Admittedly, the sentences do start getting a little more complex as you go on, but at the time a large chunk were garbage.
I definitely wouldn’t mind having a pre-made deck of sentences but there probably won’t be one that’s great for everyone. I even recently had a look at AJATT’s own “My First Sentence Pack” and that didn’t look too great: many of the sentences were repeats of others but with maybe だ changed to です (or similar minor changes).
That plugin looks pretty interesting. It seems similar to how LingQ.com works, if you’ve ever used it: as you read texts, you mark words that you know, and then when you look for new texts, it shows you how many words are in it that you haven’t marked as known (as a rough judge of difficulty). Doesn’t work *too* well for Japanese though, the lack of spaces causing difficulties with parsing of words.
I think I’m definitely going to start making a sentence deck now, even though I’ve not finished Core 6k. It’ll take a while to get enough sentence cards such that reviewing both decks will take up too much time; I might even have finished Core 6k before I get to that point :D
Ever thought of doing the “Massive Context [Cloze] Deletion” stuff that AJATT talks about these days? Apparently 10,000 sentences “is dead” and MCD is so much better, but I think I’d rather do sentences :P
> handmade decks are better
For me what puts the Core collection over the top is the collection of professionally-recorded audio clips, one for almost every sentence in the collection. Listening comprehension is a central part of what I do with the sentences, and in my view the added value you get from a custom deck isn’t worth the amount of time or money it would take to build.
Based on what I have heard, the latest tool for people who want a custom deck with text, audio, and even video extracted from TV shows is something called subs2srs. The idea is to use the time cues embedded in subtitle files to cut up the audio and video into chunks you can practice. When I tried it, it didn’t work well enough for me to switch, but perhaps it has improved since then.
http://subs2srs.sourceforge.net/
> MCDs are better
I’m doing something like that with Kanji practice, where a single kanji is removed from a sentence, and you have to write it down by hand. The Kanji practice cards are ordered just before the particular Kanji+reading combination comes up in the sentences, so they help ease into the new material. I also put some etymology information from kanjinetworks.com on the cards, and that sometimes helps. But later there will be a card for writing the entire sentence by hand, and to avoid having too much overlap I usuallly shelve the kanji practice cards farily quickly.
Overall I think MCDs do a good job of isolating one individual thing you want to learn, at the expense of introducing a whole lot of duplication. If there are 4 things you want to isolate in a given sentence, that means you need separate 4 cards for that sentence, and you may end up over-repeating it if your review schedule isn’t carefully planned.
Yeah, that’s a benefit I meant to mention, the fact that the sentences come with audio. However, lack of audio for hand-made decks is fine if you follow the other half of AJATT, that is, listening to lots and lots of Japanese as well. There’s no way I’ll be able to do that as hard core as Khatzumoto did it but I’m definitely trying to listen to more Japanese audio of late. To be honest, I actually don’t like the audio for the sentences included with Core 6k: since they’re reading them off a script, lots of the sentences tend to sound the same, kind of unnatural (especially by one of the female voices). I have the audio turned off for my Core 6k, also partly because it gets irritating hearing the same sentences over and over again (or cut-off sentences when I answer a card before the audio finishes).
…It sounds like I’m trying to convince you to stop using the Core 6k sentences but I really don’t mean to XD I’m just trying to say why I’d rather not use them. Adding all the sentences myself is definitely going to be a chore though…
What you’re doing with kanji is pretty much how AJATT describes it, from what I’ve seen (though MCDs aren’t *solely* isolating kanji).
You must be logged in to reply to this topic.