For Mandarin, Made HSK Freeware, Making More Freeware

[the.yangist]

Man, this thing took me what feels like forever. I figured I owed this site a lot of thanks for showing me some of the links that I ultimately used to create this thing, so I've decided to post my alpha version here to see if it catches any bites and if there are any higher end programming sorts who could advise me on programming the thing in a better language.

It stands now as a not-so-small freeware Excel application that allows for drilling of Chinese characters and HSK terms. The cool thing about this one I've been designing is its ordering method that improves the efficiency of drilling through use of frequency tables and such (and with later updates, Zhongwen.com's Zipu mapping) to ease the mental insertion of all of the major characters.

Anyway, here's the link. The sheets in the workbook are pretty self-explanatory, so there's no need for long directions.

http://d01.megashares.com/dl/d82a52c/HSK Drilling Public Alpha 1.xls

Also, I'm working on other freeware projects that could benefit from user submissions, so if you're willing to help with that, check this dump out:

http://sites.google.com/site/hanyuxueshi/

Let me know what you think, and thanks again for doing a lot of the early hunting for me.

Edited May 11, 2009 at 05:48 AM by the.yangist
My site returned a 404.

[roddy]

Did you get frequency tables for HSK vocab items (rather than just characters).

How does this compare with the likes of Anki and ZDT offer. Both are free flashcard apps for Chinese (or other languages) and have various decks already available - if you can describe it in terms of those two programs, we'll have a much clearer idea of what this actually does.

(your site returns a 404)

[the.yangist]

Sure. There's a lot to cover, but I'll start with differences between what I've been making (I'll call it HSKD) versus ZDT for the most part.

1. ZDT's system trips over its own complexity. There are no means to simultaneous selection of the definition training and Pinyin recall, and the way one marks one's progress appears to be well-paced, but inefficient because its definition of "progress" is based on number of instances of retention. I've redefined progress through the HSK vocabulary system in a very different way. Instead of using streaks or percentages to "drop" flashcards, one's sustained progress determines one's position on a vocab series that is already ordered by difficulty, which I defined under a few factors: stroke count, individual character frequency, and (later) Zhongwen.com's Zipu family position frequency. The efficacy in the list I've developed lies in the way in which it is ordered.
   
2. My HSKD, therefore, is not ordered by HSK level or by the user, and so he only has to focus on the accuracy of his entries while simultaneously building a vocabulary of varying levels. The system is actually built around character recognition via one's selected IME input (I created a small Zhuyin Fuhao - Pinyin trainer for my Taiwanese friends, too, if that interests anyone). This work focuses on lexical and phonetic recognition first (assuming you use Pinyin or Bopomofo), while definitions are learned passively. The point of this software is really to be able to recognize all of the terms and then be able to read a book of one's choosing and develop one's skill while immersed in the language (or at least, that's why I developed it).
   
3. Instead of you choosing the complexity yourself, my Skill Assessment makes pretty accurate guesses to the number of characters that you already know, as well as the level where it would actually be challenging to start. Your right and wrong answers shift you up and down the vocabulary list, and one's overall score increases as one learns new characters. Actually, progress on the HSKD works in a way such that one's progress decelerates as the entries become more difficult, and in general you're forced to keep track of about twenty or thirty terms at a time in order to make much advance, which sounds like a lot, but really isn't.
   
4. The text on the HSKD is much bigger. Actually Mnenosyne is the worst culprit of tiny text, not the ZDT.
   

My problem has always been that, at the end of the day, flashcard drilling is flashcard drilling, and it still has all of the same learning holes that would allow for people to make false progress. I aimed to make the learning a bit more organized and less error-prone than that, and the applet I made is the end product of that effort, as well.

There's not much more to say without going on at length about the syntax and semantics of natural languages, so I'll cut it short here.

One last thing: Some first users have been telling me that the download restricts access to the entry areas. If it does, the unlock password is shang4de2wu2wei4 (for all you Laozi readers out there).

[chrix]

Hi,

I tried your drills, but couldn't make sense of them at all.

- whatever I entered in "Hanzi drilling", it wouldn't recognise anything I entered as correct.

- whatever I entered in the other sheets, it would also lead to division errors...

any ideas as to why these errors occur? Would be glad about explanations how to use those drills correctly...

Thx

[the.yangist]

I have a fairly decent grip of what's going on.

In the Hanzi Drilling area, you'll presently get a #DIV/0 error if you enter the first character incorrectly. It's pretty much a mathematical error that results in the early stages, and it's something I haven't perfected (e.g. how to score and assure progress in the Drilling areas). I've gotten a bit closer, though, and I'll fix that bug with the new edition.

The drill system is pretty easy. You just enter the characters as you see them to the left in the topmost open light green space. The drill assumes that you use Bopomofo or Pinyin as your IME. If you enter anything else, you'll produce an error, though it should only lower the score.

I think some of my protections managed to harm proper function of the program, so I'm going to have to reconsider how I organize the sheet in the released version.

I'm redoing the entire scoring system so that the math works more easily, but offers a similar challenge all the same.

[chrix]

hi thanks, I thought you had to enter it in pinyin, but after I started entering it in characters, it was working fine. Thanks.

[skillphiliac]

I don't quite get what I should be doing. I am using OpenOffice, and whenever I try to type something anywhere, I get the error message "Protected cells cannot be modified.".

Could you give me some advice?

[renzhe]

I don't know if all the scripting done in Excel works well in Open Office Calc. This could be your problem.

[skillphiliac]

Yes, evidently. But if not so, there could be a solution...

[the.yangist]

I originally protected the document out of interest for the user so that he wouldn't inadvertently type in some unwanted space and then really be up a creek trying to undo the typing error. It's not the biggest deal, just one to consider.

I need to do more research on how to make those protections safer and more compatible with users, so my next update will be much friendlier.

For those using this alpha version, you can always get out of the protections with the following password: shang4de2wu2wei4 . Just be careful where you type!

In somewhat relevant news, I removed the Hanzi Drilling from the new update since it defeated the original way in which I intended users to acquire character knowledge.

[the.yangist]

I've gone through a fair bit of work getting this thing operating, doing about one month's work nearly full-time to get this version released. This is basically a shell, and given some tinkering on any corpus, I can make language drilling for any language.

I hope you like the fancy title!

Here are the main improvements:

1. I'm no longer using the HSK word list, but instead a word frequency corpus that is more modern, so words like "online" (網上), "website" (網站), are there, plus terms in more modern usage as of the 2000's.

2. The math is more sophisticated, and I took advantage of Zipf's law in the scoring and progress markers.

3. I opened the security for people who want to doctor with the corpus more than I already have.

4. I implemented a macro that stores your progress more efficiently, so you don't have to keep track of your position in the corpus as you go along.

I've just started a site called "Polyglot", which will basically feature all of the materials I need for tutoring and independent language learning for all of the ones I know or study. (It's bare bones, right now, as I've been paying the most attention to the applet above.) It's a Wiki page, though, so I'd love to get some input on that project, as well.

I've also got a load of ideas up the pipeline, including a Chinese etymology and affix structure, as well as some time-consuming updates for the Corpus Drilling I've made here.

[the.yangist]

One more update:

Corpus Drilling Public Alpha 2

The macros will not work in OpenOffice, as another user has confirmed.