ファブリス
Administrator
From: Belgium
Registered: 2006-06-14
Posts: 3699
Website
I'm working on adding example words to the flashcards.
I'd rather keep the layout as simple as possible (in fact I believe it would be more efficient to have just one example word per card, but that can always be added as an option).
With this in mind, I wonder if it's important to distinguish whether the On or Kun reading is being displayed, for each word?
Sebastian
Member
Registered: 2008-09-09
Posts: 520
Nice idea! I'm already using Rikaisama when reviewing so I can see how useful such a feature would be. Besides, it would be a nice addition to silence RTK naysayers.
You could use hiragana for kun'yomi, katakana for on'yomi and periods or dots to separate the readings corresponding to each kanji.
For example:
人目：ひと・め
名目：メイ・モク
About how many or what words to display, it probably would be too complicated to implement, but it would be even more useful if you could display a list of words and then check the ones you want permanently displayed on your card. If the number of people who have chosen to display each word could be seen by the side of each word too, and words could be ordered by "popularity" (similarly to how kanji stories are ordered), that would be awesome.
Just my 2 cents.
ファブリス
Administrator
From: Belgium
Registered: 2006-06-14
Posts: 3699
Website
Sebastian wrote:
You could use hiragana for kun'yomi, katakana for on'yomi and periods or dots to separate the readings corresponding to each kanji.
For example:
人目：ひと・め
名目：メイ・モク
Sounds good. What do other people think about the separators? For technical reasons I'd rather not use them, as it will make the layout more difficult (will take more space). On the other hand I was planning on using colour to emphasize the individual reading of the current kanji.
Sebastian wrote:
About how many or what words to display, it probably would be too complicated to implement, but it would be even more useful if you could display a list of words and then check the ones you want permanently displayed on your card. If the number of people who have chosen to display each word could be seen by the side of each word too, and words could be ordered by "popularity" (similarly to how kanji stories are ordered), that would be awesome.
That's what I had in mind.
I'm going to do this incrementally though so for this first implementation what I've got currently is one random Onyomi example word, and one random Kunyomi example word; both selected from "priority" entries in JDICT. So some kanji have no On, some have no Kun, and some have no words at all, which I think may be better than burdening oneself with uncommon words.
This may be appropriate if I add this first to the non-SRS mode. So each "cram" session can show up different words.
If I use the first results I can sort them by priority but they won't necessarily be the most common or logical examples because there is no manual selection yet. Hence random.
That's a step up from the "Vocab Shuffle" I did, because here at least you're supposed to know at least one kanji in each word.
bertoni
Member
From: Mountain View, CA, USA
Registered: 2009-11-08
Posts: 255
I like your layout a lot. I probably wouldn't add dividers. They're more stuff to parse, but I don't think it's a big deal either way. I guess using katakana for on-yomi is reasonable. I personally might use a different color, instead, but the difference is small, in my opinion.
Last edited by bertoni (2012 May 25, 3:56 pm)
ファブリス
Administrator
From: Belgium
Registered: 2006-06-14
Posts: 3699
Website
Nukemarine wrote:
If you can use the Core 2k/6k/10k list as the primary words and English definitions, I think it'll be better than the lengthier definitions given in the current sample. With that, it won't look as cluttered if you used 2-3 kunyomi and 2-3 onyomi examples.
The database can even have priority based on CBs vocabulary frequency results. This'll ensure a more common sample word is used.
Please note as Sebastian guessed, in the longer term the plan is to let the user pick what words they want. Once users can pick their example words, lists can be made to represent Core 2k, or anything else. JDICT is provides a common, exhaustive set, with "preset" definitions that save users time.
Technically speaking the definitions displayed on there are a concatenation of several glosses. The first gloss (should be the one the most in use) would have been "(adj-na,n) (1) (uk) absurd; unreasonable; nonsensical; preposterous; incoherent;".
I purposely selected long definitions to see how it would fit in the Photoshop layout.
HelenF
Member
From: UK
Registered: 2012-04-11
Posts: 31
Ah, what I've done is probably similar then.
I don't have much idea of how many of the words can be solved with an algorithm, how many others have uncommon variants that could still be argued to be individual kanji readings, and how many don't have individual kanji readings at all. I suspect quite a bit of manual checking would be involved in getting a good answer, so a central data table e.g. in JMdict would be a useful thing.
I think people were talking about using Mecab to generate furigana, but that works for morphemes, not the individual kanji. Split furigana appear in some printed material though.
I'll search some more.
