Feeback Suggestion: Replace %New Words with %Known Words

I would like to make the Suggestion of changing the counter for %New Words(Blue Words) to %Known Words(White words).
The reason is that I think the %New Words does fail at reaching it’s goal, which is indicating to the user how easy or hard a lesson will be.
Lets say I have a lesson where 10 Words are New 10 Words are LingQs and 80 Words are known. It would display as 10% New words informing the user that this will be an easy lesson where most of the text is understood.
However a text with 10 New, 80 LingQs and 10 Known words would also show 10% new words but it would be a really hard lesson as 90% of the words would be unknown.

Another drawback is that %New words is the only Sort option when searching in lesson collections or in themes or in the library. I can only filter by %New Words but the LingQs are most of the time also words that I don’t know (unless they are level 4 or maybe 3)

Also there is the disadvantage that a text that I have already read will be marked as 0% new words. Therefore for every lektion that I already read I don’t have a convenient way of telling how easy it will be.

TLDR: I would like if the white words (marked as known or ignored) to be the primary indicator instead of the current %New Words.

What do you think ?

2 Likes

This is a long response. Be forewarned.

Just as with your example of 10 New, 80 lingQs, and 10 Known. This is surely harder than 10:10:80 Known, but still easier than 80 New:10:10! Of the 80 lingQs in your example, as a lingQ could be at level 1, 2, or 3 (and 4 too?), at the very beginner stage, you have lots of 1s, but by the time you are maybe an upper beginner/lower intermediate, you’ve slowly moved up your lingQs and a lot of those lingQs are actually level 2 and 3. You have encountered them before, likely several times. Personally, I have a lot of words which are still lingQs, but when I read them, I don’t look them up. I either understand the sentence (which doesn’t require me looking up the word so I move on) or I get the word correct and move it up one level (so I have to get a word which I’ve lingQed correct 4x before it’s Known).

As a quick example, I just went through a few pages of a lesson and encountered 59 lingQs. Of those 59 lingQs, I looked up the definition of 9 of them. So that’s 15%. Now let’s compare this to New Words. To do this we look at the following for the last week: Known Words - lingQs Learnt = 67 (technically this is probably higher because lingQs Learnt includes phrases, but usually there aren’t too many phrases). 143 lingQs made. This means that 143/(143+67) = 68% of looking up New Words and lingQing them (technically probably a little lower because there are a few New Words which you bin).

As you can see, a quick calculation shows that for me (everyone is different and it also depends on what you’re reading, etc. but this is a quick ballpark figure), there is a significant difference in the amount of times I look up a New Word compared to a lingQ, 68% vs. 15% in this example. And then consider that it takes much longer to encounter a New Word than a LingQ because you have to look through multiple community definitions or write your own definition, compared to simply looking at the definition. So this means that New Words have a larger impact on both how challenging a text is and how long it takes to read than lingQs do. Though, perhaps you are right in that level ‘1’ lingQs still may have a high look-up rate and are maybe worth considering.

I mentioned this several months ago, but a way to look at “how hard is this lesson?” can be seen as “How often do I have to look a word up in the dictionary?” (as vocabulary is usually the biggest obstacle in understanding the text). We have the raw number guess (these are New Words, but you minus a few percent for proper nouns and depending on your level in the language you minus some for encountering various forms of words you already know). But this does not take into account how long the text is. If you have 50 New Words for a text which is 200 words long, that’s a very hard text. But if those 50 New Words are over a 2,000 word text, it’s a very reasonable text which you should be able to read fast.

This is why I proposed a few months ago to have instead the metric ESTIMATED LOOK-UPS PER 100 WORDS = UNIQUE NEW WORDS / TOTAL WORDS * 100.

In this metric, if we have 50 New Words/2000 Total Words*100 = 2.5. This means that after minusing some for proper nouns and for words you already know, you are probably looking at having to consult the dictionary maybe 1.5 times per 100 words, which would be very reasonable for quick reading.

But as to your point, maybe this could be altered to the metric of:

ESTIMATED LOOK-UPS PER 100 WORDS = (UNIQUE NEW WORDS + LEVEL 1 LINGQS) / TOTAL WORDS * 100

or alternatively you could add some weights if you want (it takes 3x as long for a New Word vs a lingQ and you look up 70% of New Words vs 30% of level 1 lingQs):

ESTIMATED DICTIONARY LOOK-UPS PER 100 WORDS = (3 * UNIQUE NEW WORDS * 70% + 1 * LEVEL 1 LINGQS * 30%) / TOTAL WORDS * 100

But from this, you can see that the constant for New words is 2.1 vs 0.3 for level 1 lingQs. With this order of magnitude difference, you might as well simplify it and go back to the original equation, as it’s an estimate anyways.

And I mean, if you really want, you could do the opposite of this equation (Known Words + lingQs) / Total Words * 100 and be able to get the same amount of information out of it, but ‘estimated percentage of a text you understand without using a dictionary’ just isn’t as meaningful as ‘estimated times per 100 words you need to use a dictionary’.

TL;DR I think New Words is the most relevant variable, but I think the metric should be altered ever so slightly to include the word count of the text: ESTIMATED LOOK-UPS PER 100 WORDS = UNIQUE NEW WORDS / TOTAL WORDS * 100.

1 Like

Of course a ESTIMATED DICTIONARY LOOK-UPS PER 100 WORDS would be better than both %new words or %known words. But it would be extremely hard to implement. So it is not realistic that we get this.
It may also depend on how experienced the user with a language already is. If you only lookup 9 out of 59 lingqs it means you know most of them. But for me that is not the case. I have 3817 level 1 or 2 out of a total of 4719 words in my vocab list. My total known words are 2071.
And currently I am at a point where there are no more texts in lingq with less than 40% new words. But also a bunch of Lingqs. So every text has way more than 50% of words that I have to lookup. So I am kinda stuck on rereading old material.
But there is the problem that I cannot easily filter for texts that I have read that are easy. As those are displayed all with 0% new words, but range in a difficuilty from 95% known words to 90% level 1 and 2 lingqs.

But maybe instead of replacing the %New words it would be a better option to add %Known words as a Second counter as well as a filter option. That way we would have both options.

1 Like

Of course a ESTIMATED DICTIONARY LOOK-UPS PER 100 WORDS would be better than both %new words or %known words. But it would be extremely hard to implement. So it is not realistic that we get this.

Nah, it won’t be hard to implement. We already have Unique New Words (those are the blues words) and we already have total words for a course (which is obviously a sum of the totals of every lesson). So we already have the two variables. The metric just needs to be changed.

And currently I am at a point where there are no more texts in lingq with less than 40% new words.

So this is the issue you’re trying to solve. Obviously an important issue. Maybe you need to look for your own material, if the Ukranian libraries are limited? Also, if you have already gone through the lessons and you are wanting to review them due to lack of material, surely you are open to restudying lessons which you previously completed, whether they are 80% lingQs or 20% lingQs, no?

1 Like