The Chinese Mini Stories have so many errors.
Here is just another one - see a doctor is not one word, it’s two. And from my other post, there are so many pinyin errors which I reported - no action was taken.
Is it because I paid already and that’s the end?
Since I presume nothing will happen - how can I edit or correct the pinyin?
You can’t edit it on your end. I’ll ask our content managers to look into it.
Looking at it will not help much. Fixing it would.
Really, I have paid for LingQ and now I find out that it has lots of errors. And even after reporting they don’t get fixed.
I don’t expect computers to be as smart as humans. They aren’t. LingQ relies on the fact that languages have a space between words. But Chinese doesn’t. In Chinese, you need to really understand the sentence to split it correctly into words. Computers can’t do that. Even humans get confused sometimes. I get confused by proper names.
I first evaluated LingQ in 2022, for my Chinese study. This problem was one reason I didn’t use LingQ for that. I reported the problem, but what can they do? If someone invents an AI program that solves this, I’m sure the LingQ folks will try to get it.
In 2023 I started studying Turkish, and found that LingQ worked will for me. I have been using LingQ for a year, for Turkish study.
Fixing it, maybe?
Has nothing to do with AI. I don’t talk about stuff that I import, it’s the lingQ provided Mini Stories - and sorry, they should be 100% correct.
This is really pathetic. There is already an error in story 1a.
On top, it’s all known that auto translations have issues, hence you need full editing capabilities for material that you import.
At this point I can’t recommend LingQ for Mandarin Chinese. The errors are too distracting.
You can click on the three dots up right and then on print lesson to copy the entire text, from which you can then create your own lesson. There you will be able to add whitespaces to your heart’s content. The audio file can be downloaded as well.
I tried with Mini Story 03 - same problem… 着 > zhao is wrong, it should be zhe.
着 has 4 pinyin options. Seems the zhao is hard coded to one sound only. The AI spoken story gets is right though.
I haven’t tried fantasy words (看医生 - there are a few more, not too many), there it may work.
I would second what @Qulla said. Even if there are no good algorithms for word splitting in Mandarin - which I wouldn’t simple assume, their Japanese word splitting is also much worse then the one on my Kindle, which isn’t even aimed at a japanese audience mainly, and the Japanese dictionary does a much better job, too -, the Mini Stories are custom made. One of the reason a new language has to be added officially is that they want to make sure that those stuff is okay, they say, so providing them mostly error-free is the bare minimum.
The actual problem that I see is that LingQ tries to force one method onto all languages, no matter the differences. Wouldn’t it make more sense if the user could specify what a word is, so instead of selecting one out of several (often wrong) predefined chunks, the user do so himself. Select a couple of characters that you consider to be a word and create the definition. We already have the ability to turn several subsequent words into a lingq, so it is not much of a change imho.
I think having one story in different translations is a good idea. In case you start a new language you start with a familiar story. That’s good.
No, because in 98% of cases the AI gets it right. BUT, an edit functions would be great!
This said, the Mini Stories have a report function. Let’s see how it takes them to fix:
1a:
pinyin over 着 should be ZHE, not ZHAO
22: This are NOT single words:
感冒药 > correct is 感冒 + 药
看医生 > correct is 看 + 医生
Lets see how long it takes to fix it. Today is 12th Nov. 2024
That isn’t what I have meant. What I was referring to is the way the system works, working on the assumption of an european language. For those it works sufficiently well, but once you step out of that area and use it for a language that works by other laws, like not using space to separate words, using different signs and characters alltogether, a non-phonetical script etc., you get quiet some downsides.
For example, I am learning Korean. It uses a phonetical alphabet, so I don’t have the issues that learners of Japanese or Chinese languages have in regards to pronounciation. However, it is a highly agglutanive language, so each word appears in dozens of different forms. And in difference to Germanic languages, for example, the word stem doesn’t change. So once you familiarized yourself with the grammar, you will recognize those different forms representing the same semantic and you will know what the modifications made to the word mean.
But still, in the worst case when there is neither a good popular meaning nor a good ai translation (which is complete nonsense half of the time), you have to either type in the same definition over and over again, or just create a few lingqs and ignore the rest, making the whole known words statistic pointless. I don’t care much about the latter and can type pretty fast, but still I consider this to be something that slows me down. It is just needless, and there are algorithms that can pretty reliably identify the correct wordstem (I mean, I can, and I am still pretty bad at the language). So why do I have to do something that might make sense for a Germanic language, but is completely pointless here?
And in a different text that I read later on I may encounter those versions of the word that I have not lingq’d, so I create a new definition again for a couple of those which may differ from the first on made. And later on in yet another text I may encounter both version and than have to decide whether both make sense or the older one is faulty due to my lack of language skills back than, if only I knew which one the older one is. And if both make sense, I have to either do quiet some copy and pasting or live with the incompleteness of the single definitions. Fantastic.
