fair-rose•2w ago
Is there any way to resolve cases where the parser gets the wrong reading of a word?
As the title suggests, is there any way to resolve cases where there’s more than one reading for the same kanji and the wrong one is the one being picked up by the parser? Especially in cases where it picks the clearly less common word
In the following example he uses 味 (あじ) and not 味 (み) but the parser insists on picking up the second option. I checked Yomitan and it picks the first one which makes sense.
Cases like this can get a bit annoying cuz I get a ton of recommended sentences that aren’t actually anything and vice versa.
14 Replies
Send it to the card creator and in the target word field turn on modify syntax and you can change the reading there
eager-peach•2w ago
Cases like this can get a bit annoying cuz I get a ton of recommended sentences that aren’t actually anything and vice versa.If you know a meaning / reading for the character just mark it known. No need to leave it unknown so that it messes with sentence recommendation
correct-apricot•2w ago
Apropos this, I guess it wouldn't be feasible to have the word audio take the modified syntax into account, right? For now when this happens I usually just temporarily type the word in hiragana in the target word field, re-create word audio, then go back to the kanji version
But that can sometimes result in wrong pitch
fair-roseOP•2w ago
I know yea but the issue here is that this is a word I already know, doing that would be creating a card for a word I already know. My problem is more so it detecting the wrong reading in the first place
Hmmm ok I guess since I’m a beginner it gets a lil scary to mark a different reading as known when I in fact do not know it but i understand where you’re coming from since I’ll most likely surmise the meaning from the kanji itself.
The parser isn’t perfect, it just sometimes get the readin/word etc wrong. If you already know the word then there’s no need to make a card and you can just not set it to anything, just ignore it as a parsing error
fair-roseOP•2w ago
Still it’s a shame and kinda weird that it’s picking up not only the wrong reading but the less common one for that kanji I feel like that’s should and could be easily fixed
I’ll most likely do that yes
I think it’s based on the order it was set to in the dictionary they’re using 🤔
eager-peach•2w ago
You're not marking a reading as known, you're marking a character. There's literally no downside. Your known word count is only a rough approximation anyways and its benefits are motivational (tracking progress) and functional (comprehension score and rec. sentences)
They don’t make the dictionaries from my understanding they make use of premade dictionaries I believe
fair-roseOP•2w ago
Likely yea which I guess in this case compared to yomichan is different
I would assume so as well
Am I? Because I tried pasting the character by itself and then the phrase in question and the character by itself is marked as known but when used in that sentence it’s marked as unknown
fair-roseOP•2w ago
eager-peach•2w ago
Well it's gonna depend on the sentence. More precisely speaking, you're marking whatever the parser considers to be a word given the sentence.
fair-roseOP•2w ago
Hmmm okok I’ll try to ignore these cases for now and see if that annoy me a lot especially in cases where the reading is in fact very different in meaning. Thank otherwise I’ll start marking them as known as well
Thanks for the help and answers! @HulK @Minoqi 🇰🇷 🇯🇵 🇨🇳
eager-peach•2w ago
sounds good!
gl with your learning