I dont understand these numbers. I have never seen anybody on Lingq who has reached 280.000 known words. The highest numbers I see are around 120.000. And surely by the time you have reached 280.000 known words, your read word count is going to be in the millions and not at 112000.
112.000 words are only about 40 bookpages. How would you learn 280.000 word forms reading only 40 pages.
Also I dont think that 9.000 word groups make 280.000 word forms. That would mean that every word has about 31 forms and that you have encountered them all.
Am I to stupid too interpreted your numbers? What do I not get?
The formatting is a little hard to read…but 280,000 known words = 11200k read word count = 11,200,000…not 112,000.
At least that’s my take.
As for whether 9,000 word groups make 280,000 word forms. I don’t know exactly. Obviously it depends on the language. Also, do they count word combinations…like in German to make the various tenses you have helper verbs, but the main verb may stay the same across all the first person, you, third person, we, etc…
I haven’t looked at the article to see what they say about all that.
Thanks. I understand the read words count now. This makes sense. But the 9000 word groups to 280000 word forms is much to high. I always thought that the numbers Lingq gives you are much too low. 35k or so as level 2 advanced but using this conversion rate of 1 word group= 31 known words, then 35k known words are little over 1000 individual words. That is hardly an A2 level. .
My intuitions as a czech learner, a language with plenty declensions, is that 280k known words should be close to 20k word groups. Especially since you do not have to encouter every word in each form to know the word and understand the patterns. It seems to me that 280k words is already very, very advanced and I checked the top achievers of other languages here on Lingq and never saw a number like this. So has nobody on the entire site here ever reached a C1 level of around 10000 word groups?
I think the numbers are a projection based on the numbers of the article and how lingQ calculate words. But the data won’t be easy to calculate at all because higher are the numbers different (probably much lower) are the amount of words for each word family.
I believe the cross-data might be useful until 5k/6k then the lingq’s numbers would probably be much much lower.
Have a look at regler with 272 000 word forms and 8 500 000 read words.
The connection between known and read words I made based on my experience (read 40 words to learn 1). As you can see above regler’s ratio is 31.
The values from 1000 to 9000 words are the most frequent 9000 words in the language, not the exact number of words that the learner knows. All language learners acquire thousands of words that are outside of the most frequent words.
I, for example, know roughly 13000 english words, but lots of them come from metallurgical field and other areas, and they are very rare, so they will not count to the top 9000 words.
For Paul Nation it’s important that people learn vocabulary useful for daily life, that’s why he talks about the popular words.
Quote:
The table shows that 35 000 word forms equal to more than 4000 known words (see row 4 with 94,2% comprehension). The conversion rate changes when learner reads more and more. At first the rate is 5, i.e 10000 word forms equal to 2000 words. Then it drops to 31, while after reading millions of pages the chances increase that you encounter all of the word forms for the words from the top 9000.
From my experience, I’d have to disagree with some of the math that’s represented here. I think the ratio of “LingQ words” to dictionary words is a lot lower than the above table would suggest.
One big problem with running comprehension numbers with German is the prevalence of compound words. You could import text into LingQ and the app will tell you you have 15% of unknown words, but if all of those are compound words that you already know the components of, the text could be 100% comprehensible to you.
I decided to graduate from LingQ with German at 50K known words for mostly this reason. At that point, I felt that most blue words I encountered were either compound words, proper nouns, or obscure, low frequency words. Only a small percentage of words encountered after 50K were what I would consider “unknown words of value.” (My 50K known words did not include proper nouns, which I tend to ignore in most cases)
I simply have to disagree with the 280K LingQ words to 97.9% comprehension. At 50K known words, I was able to read a Kindle book just fine, with only stopping to look up less than one time for each 100 running words.
With French I graduated at 40K and had about the same lookup ratio at the time.
After reading these interesting posts, I decided to check on my statistics, something I don’t normally do. After more than a year on Lingq I have accumulated ~37,000 words read in Norwegian. Huh. Seems kinda low for the time I’ve put in. The number doesn’t even seem to be in the right ballpark.
So I recorded my starting data, read a lesson as I always do, then looked at my data after. I gained several new known words, and a couple dozen or so lingqs, but my words read did not increase at all.
Then I tried doing something I rarely ever do. I clicked ‘complete lesson’ instead of just x-ing out when I’m done. That did it. That increased my words read count. I’ve been skipping ‘complete lesson’ at the end because, to save time, I don’t bother manually ignoring all of the words that I want to ignore, and I don’t want lingq to mark them as known, so I just close it instead. I do the same thing when I reread a lesson-I wonder if reread lessons should count in the cumulative total?
After more than a year of doing this in all of my languages here, I’m now aware that I’ve lost something valuable in my statistics. Wish I’d have figured this out at the beginning. Now I guess I’ll never know what my real words read counts are.
yes, that’s true. But you can also see that at the end, the difference between 55K and 280k words is only around 2% (95.5% - 97.9%) in comprehension. Which make sense to reach around 50k/60k known words and don’t bother anymore, even if the hypothetic 280k words stats are correct. It doesn’t add much value anyway.
My stats say I have not much more than 3000 words in French, but I can watch the French news and understand 80-85%. LingQ doesn’t count the hours of listening/reading I do elsewhere on youtube, podcasts, books. Listening does one thing, reading does another thing, but speaking is an actual active skill you must practise to master. I have little time for accumulating certain statistics because they often tell you very little. Like those people with a 5000-day streak on Duolingo who can’t really speak any of 25 languages they are ‘learning’.
The problem with not being able to understand everything, especially in material from native sources, is because things like books and TV and films use vocabulary that a native speaker has acquired merely by dint of being born into the language and hearing it all the time because it is the only way to communicate. Also many of these words are not ‘everyday’ words, they are carefully-chosen synonyms that probably wouldn’t be used in off-the-cuff speech. I don’t worry about it. My core vocabularies and whatever I add to them will always lack words which turn up. Even native speakers don’t know many words used in books/films etc.
So yes, it will sort of ‘click’, but this will happen more than once and in stages. There is no great reveal, but the mist gradually clears.