Lately LingQ seems to be struggling with Japanese more than normal.
It’s attaching particles to words which is increasing the amount of unknown words, also leading to inaccurate word count, which means you either have to edit lesson, manually separate the particles then choose how well you know the word, and if you already know the word you can choose the ignore button(assuming you know you’ve come across the word in LingQ before).
This is time consuming and I feel undermines what makes LingQ so good. This seems to mostly be a problem when importing from YouTube.
Word splitting is very inconsistent in Chinese as well.
For example, every combination of [number][classifier][word] shows up as a separate “word” in LingQ, even within the same lesson.
This means that this simple “one person” phrase generates six LingQs:
一个人
一
个
人
一个
个人
Then, changing from “one person” to “two people” gets you another 3 LingQs. Changing that to “three people” gets you yet another 3 LingQs. And so on. The same thing happens with “some people”, “more people”, “fewer people”, etc.
This leads to a massive explosion of unknown words (and known words) in LingQ that correspond to a relatively very few real words in the language.
I’ve accepted that this is just something I have to deal with in a language that does not use spaces between words.
That said, if LingQ could recognize even that apparently simple pattern above it would do a lot of good.