Frequency in what? Light novels? Manga? Newspapers? Pro wrestling magazines?
I would assume we all understand that there is no perfect frequency list. And that there are some generalities in my post. But it is possible to build lists that encompass the various sources. For example for the work i do in Foreign Policy and News, I would have a slightly different list than say a light novel reader. But there is going to be a great deal of overlap between newspapers and light novels. Probably less so with wrestling magazines. I am going to see æż (politics/government) far more than a light novel reader but you will also see it. For me it would probably be in the top 100 or 200, for you it might be 500. Also all of the frequency lists i have seen look pretty similar give or take a kanji here and there
In clown magazinesâŠ
Sorting by frequency makes the most sense, sure, but I think two pitfalls are going by unnoticed
-
Frequency is highly context dependent past the first couple thousand words. And the further you go, the more it matters.
-
The real important part for reading isnât kanji, itâs words. So the goal shouldnât be any â% coverageâ of kanji because that doesnât really mean anything. They would need to focus on words, which would probably mean kanji are introduced only as they are needed.
I was going to post this a while ago, but the Kana only threads seemed to disappear from my feed so I left it. Since theyâre coming back, hereâs my thoughts:
BTW, the tl;dr version is I am against the kana only additions, but I also hate the âpeople should already know thisâ arguement.
Iâm currently doing a BSci in Astrophysics and Computer Science. I just finished my final exam for a third year CS programming unit (C++). When that unit started talking about variables, it began with what variables are, how intâs differ from floats, how computers donât natively know strings of characters etc. Now not only is this a third year CS unit, but you have to have done two programming units as prerequisites to qualify for this course. Therefore explaining the basics of what a variable is is simply a waste. At that point youâve already been required to know the IEEE standard on floating point variables, so why does this unit begin with describing at a basic level what they are?
So Iâve been e-mailing one of my professors back and forth about some research and I mentioned this. The bulk of the e-mail was about something completely unrelated, so he only gave a one line answer to this question, which I found quite comical, but also completely summed it up. And I quote: âWhy would you start anywhere but the beginning?â
Itâs true that the WK philosophy has always been to teach kanji, and i think thatâs what it should stick with. But to say theyâre stupid additions because (quoting this thread):
Theyâre only basic for people who already know them. We all have to start at the beginning at some point. While Iâm completely against the kana additions, I think itâs just plain elitist to claim that these are so basic they donât need to be taught.
We were all taught these in the beginning
And the beginning is exactly where theyâre starting.
Yes they should be optional if they need to be here at all, but then what weâre asking for is optionality on what is currently a completely new feature. Requesting new features for something that is in itself a new feature is probably pushing it a bit.
Ok this turned into much more of a rant than I wanted. Hopefully my point is made in there somewhere.
Apologies.
âA 3rd year course is not the startâ
Setting aside the thread, that feels like a terrible response that misses the entire essence of âpicking up where you left offâ. I wonder if he doesnât use bookmarks and just rereads books from the beginning every time he puts them down and picks them back up. He probably just wanted to sound slick with a one liner, I imagine.
The more feelings-based type of posts have started at least a week after the changes, plus you have to consider the entire time it was announced in the dev forums as a api-only thing. I was half-expecting to be among your callout-ees, but I guess Iâve only hearted every comment that communicated âI feel insulted by thisâ and raged vicariously through them
100% placid civility and purely logical arguments get slowly eroded after weeks upon weeks of no response / non-responses from the WK team. Weâre at the stage of memeing and playing a little looser with our feelings now
I feel most people who donât even have any interest in Japan know âKonnichiwaâ means âhelloâ. I kick my Connie Chihuahuas with âkâ
The thing about a frequency order is that it conflicts with wanting to teach kanji A before kanji B where A appears as a subpart of B â so ideally you teach ć before ć, and so on. Sometimes you get lucky and the simpler-shaped kanji is also higher-frequency, but sometimes you donât. Heisig RTK takes a view very heavily on the âteach parts before using themâ end of the spectrum. WK doesnât go that far but it does I think try to account for both frequency and logical order. I donât think a pure frequency order would be easy to learn at all.
Love that throwback to when Vanilla suggested to swap those two because ć is not only simpler, but also more common.
Adding ăăă«ăĄăŻ took priority before trying to improve the kanji teaching process i guess.
Teaching kanji by frequency sounds good on paper, but it for sure wouldnt work for me or majority of people. If i saw è° on level 1 i would probably think learning kanji is impossible, but seeing èš çŸ æ kanji before that made it much easier to learn, despite è° being more common than those three.
WK does decent job of compromising simplicity and frequency as is, it just needs slight adjustments like the ć ć swap or maybe somewhat lowering level of some very common kanji despite their complexity.
Also by the way, 60% does sounds impressive, but it is actually so low that its unreadable anyways, plus most very common words arent usually very descriptive (like the è° i mentioned), so understanding them usually helps less with understanding the text than some rarer (but still common) kanji. By the point you reach about 90% where reading starts to be viable (just my opinion), youve seen all the most very common ones anyways.
I agree, but I donât think that means frequency is a bad idea, just that that implementation of frequency is a bad idea. What I would personally do would look at making the first 2000 words roughly match the first 2000 kanji containing words in a frequency list. Then the same for the next 2000-3000, and so on. That way you still have a good amount of wiggle room for ordering while still teaching the 2000 most common words so that people will have the best shot when they start getting into reading at level 20 or so.
This would at least save you from the current system where there are several words not even in the top 10k in the free first 3 levelsâŠincluding one on level 2 not even in the top 20k and one on level 3 almost in the top 70k that Iâve never heard of (and the first google result is wanikani for lol). I
That sort of re-arranging would mess up everyoneâs currently running SRS schedules, but from a broader perspective might be more useful than teaching absolute beginner kana-only words.
Yeah I donât necessarily mean wanikani should just shift it ä»æŽ all at once, but thats what I think the best approach is for teaching wanikani style. They really dug themselves in such a hole with the initial vocab selection and order, though. Their new stated goal is to teach for reading fluency, but I think its a very slim chance they ever end up doing that effectively honestly. But they could probably go from bad to alright with some slow minor shifts over time. In fact, they are kinda already doing that if you look at the general movement of vocabulary in updates.
Yeah, when I wrote about it earlier in this thread, as soon as WaniKani decided (?) to switch from kanji-only to a focus on reading fluency, there is N things that would need to be heavily refactored, bit-by-bit for the task to be achievable. Which as pessimistic as I am, is still achievable, but not with the current approach.
I think eventually weâll get there, but itâs gonna take a lot of pain and sweat . I could maybe wait for when itâs done and do another run of WaniKani from the beginning.
moving items in WK levels doesnât have to mess up the SRS schedules. once an item is in the review queue, WK could simply ignore the WK level of the item.
âŠwhich is what they currently do when they move items (i wasnât sure, but i looked in WKstats and see items at various SRS levels in much higher WK levels)
When was the last time a mod actually addressed any of these issues?
And yeah, I tried it, and I do not like the additional kana only words. They waste my time and pad my correct totals, so itâs more time consuming to see how well I am doing (you know using mods to show me my total percent correct as the native one was removed and still hasnât been replaced). Would at least like an opt out function or the autoburn ability as mentioned here.
I just started learning Japanese about a month ago, and using WK just over a week. I read their announcement post and I think it is a good idea as it allows me to learn more words at no inconvenience. It is implemented into the same website I learn kanji and I love this website. Someone else replied saying if you know the words so well, it wonât take long to burn them and shouldnât be a big deal. Why not answer a few easy words every now and then when trying to become fluent? You can see the fruit of your labor and how far you have come. IDK, I am only level 1.
I donât think anybody is suggesting that those things shouldnât be taught. I didnât know what ăă means before I started learning Japanese.
Some people are questioning whether items youâre going to see basically every time you engage with Japanese, or which generally arenât hard to learn, need to go through the whole SRS process. I donât think that everything should be learned through SRS.
And even more people, like you, wouldnât mind the additions if they were optional.
Thatâs specifically not what WK purports to do though, it wants to teach elementary components first to make it easier to decompose more complex characters later. Thatâs why äș is taught so early for instance. The reasoning is outlined in this tofugu article.
I think thatâs a great idea personally, and actually Iâve even complained on these forums that I think WK doesnât push the concept far enough as it is: It feels like WK somewhat loses its way partway through
For instance today at level 34 Iâve been taught ć° while èŹ was taught at level 13. I wish WK had taught them the other way around, frequency be damned, because it would have made it easier to remember the realtively complicated èŹ kanji. They also share the onyomi (something that WaniKani dosenât point out, preferring some nonsense English-based mnemonic).
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.