One million sentences to mastery?

General discussion about learning languages
Cenwalh
Green Belt
Posts: 267
Joined: Thu Mar 28, 2019 9:14 am
Location: UK
Languages: English (N), Spanish (C1), Catalan (B2).
Language Log: https://forum.language-learners.org/vie ... 15&t=12467
x 848

One million sentences to mastery?

Postby Cenwalh » Wed May 08, 2019 10:39 am

I was watching a video from deka glossai on YouTube where he states that research shows that "absolute mastery" of understanding a language (but specifically he's talking about Latin) begins at 1,000,000 sentences processed either by reading or listening.

I don't want to claim that I think this is true but I do have two questions:
  • Can anyone find the research he's talking about? I've looked on Google Scholar, but I can't see anything about number of sentences processed either for Latin or language learning in general, and it's not in the description
  • How many sentences do you think you've heard/read in your target language(s)?

As a reference, an average English novel has about 250 pages, about 250 words on a page, and 15-20 words in a sentence. So reading an average English novel might give you around 3,500 sentences. Although a simple novel would give you fewer words per sentence, so it does seem a strange measure.

Video for reference (relevant part is at 8:30)
11 x
Double SC films: 200 / 200 (updated 2022-07-28)
Double SC books: 34 / 200 (updated 2022-07-28)

Cavesa
Black Belt - 4th Dan
Posts: 4960
Joined: Mon Jul 20, 2015 9:46 am
Languages: Czech (N), French (C2) English (C1), Italian (C1), Spanish, German (C1)
x 17566

Re: One million sentences to mastery?

Postby Cavesa » Wed May 08, 2019 3:27 pm

I love this video. Before going on about why, I would like to answer your question:

Yes, I find the numbers plausible. Based on observations on myself and others on this forum, and on our super challenge, it is not unreasonable to say that the 20000 pages (the original double challenge) could be the path to mastery, or at least a really good level. That is like 250000-300000 sentences (following your numbers. a random book I've just opened confirms it. But I've also got denser books, but the differences between individual books are not that important in that amount), that is twice or three times the middle level of comfortable reading that he mentions (100000). Let's take into account that in the living languages, you also add tons of listening to that number during the super challenge, so if we counted "sentences encountered" by both senses, we might get significantly more,perhaps up to 500000 sentences for one whole original double challenge. But not a million.

So, while I may personally think that mastery in a not too hard language may require less than the million (and I believe that talent and intelligence plays some part in that, but not the main one, sure), I am definitely convinced that you cannot read a million sentences and not be awesome. Therefore, it is a wonderful goal for a dedicated learner aiming for a high level.

I am tempted to take the 10000, 100000, 1000000 numbers as my goals next time :-)

If there is any research proving those numbers, even better.
But I am tempted to post a link to this video next time I see someone complaining they are not fluent after two months of learning on the internet :-D

What I love about the video: it explains and answers various questions:

-how can Latin be taught and learnt better. I wish I had known this back in high school, I could have become a proficient reader. If I remember correctly, our Latin teacher was switching from the traditional textbook to Lingua Latina the year after my class. What he talks about is much closer to my usual learning method. I believe strongly in the combination of grammar learnt without fear, a bit of translation, and tons and tons of input. But the disbalance in favour of grammar translation without the input was too strong even for my taste back then in the Latin classes.

-"The amount of reading is much more important than the quality." Please, repeat that every day to every language teacher with the snobbish attitude, who recomends either the most famous classics or nothing at all to their students. Let's also notice that the video actually talks about extensive reading in both the sense of reading a lot and not overanalysing everything with a dictionary ad nauseam. At least not in all the phases.

-a part of that attitude is a good learning curve, created by the appropriate reading choices. The author basically gives us a very clear overview on how to progress from the easiest step to the hardest one. I love that. Something similar is actually what I recommend to other people interested in extensive reading all the time (actually, I'd love to know what the author of the video thinks about the Latin Harry Potter and the Hobbit, whether he'd recommend them, and to which phase would he add them. I guess the easy non bilingual reads. Unless he is against translated modern works, that is possible). It is beautiful to see that this really is a universal principle, no matter what language you are learning. I find it fascinating, that in languages like Latin, it basically means travelling from now to older and older works. That is beautiful.

-I think this video is a good guide for all the learners interested in living languages primarily for reading. It doesn't patronise them like "all the good learners have conversations as a priority" (it cannot, Latin is dead), it very pragmatically shows that even the listening and speaking have their place in learning to read well as useful tools, not waste of time and torture.

-Another thing I really like is the idea of progressing from passive knowledge to speaking. That is something not rare on this forum, I'd say. But in the outside world, it is almost heresy. Speaking is being considered the universal holy grail. There is the goal of speaking from day one, and people obsess about langauge practice before they can even say anything, and people in classes learn mistakes due to being pushed to production they are not ready for. I am not saying the "speak asap" idea is universally wrong. But the other option, to prioritise comprehension first (and even for a long time, 1000000 sentences take a while) with production being a minority part of learning, is legitimate too.

Really, this video has so much to tell us about language learning in general, not only Latin! (And yes, it makes me desire to return to Latin right now! :-D )

Thanks a lot for bringing it here!!!
23 x

Stefan
Green Belt
Posts: 379
Joined: Sun Dec 20, 2015 9:59 pm
Location: Sweden
Languages: -
x 920
Contact:

Re: One million sentences to mastery?

Postby Stefan » Wed May 08, 2019 5:08 pm

I have never heard these numbers but for people interested, I would recommend the research done by Paul Nation. He's done countless of studies on vocabulary acquisition, extensive reading, word families and comprehension. He goes through the number of words students and kids know, how many times you need to come across a word to remember it, how many books you'd have to read, etc. He released a book in 2017 titled "How Vocabulary is Learned" but I haven't read it yet.

With this said, I don't know how well-founded his research is.
3 x

Cenwalh
Green Belt
Posts: 267
Joined: Thu Mar 28, 2019 9:14 am
Location: UK
Languages: English (N), Spanish (C1), Catalan (B2).
Language Log: https://forum.language-learners.org/vie ... 15&t=12467
x 848

Re: One million sentences to mastery?

Postby Cenwalh » Wed May 08, 2019 7:55 pm

Cavesa wrote:Yes, I find the numbers plausible. Based on observations on myself and others on this forum, and on our super challenge, it is not unreasonable to say that the 20000 pages (the original double challenge) could be the path to mastery, or at least a really good level. That is like 250000-300000 sentences (following your numbers. a random book I've just opened confirms it. But I've also got denser books, but the differences between individual books are not that important in that amount), that is twice or three times the middle level of comfortable reading that he mentions (100000). Let's take into account that in the living languages, you also add tons of listening to that number during the super challenge, so if we counted "sentences encountered" by both senses, we might get significantly more,perhaps up to 500000 sentences for one whole original double challenge. But not a million.


I'd never actually realised that the super challenge was so big! My own li'l challenge of 5,000 pages in the next year that I've set myself seems paltry in comparison.

-how can Latin be taught and learnt better. I wish I had known this back in high school, I could have become a proficient reader. If I remember correctly, our Latin teacher was switching from the traditional textbook to Lingua Latina the year after my class. What he talks about is much closer to my usual learning method. I believe strongly in the combination of grammar learnt without fear, a bit of translation, and tons and tons of input. But the disbalance in favour of grammar translation without the input was too strong even for my taste back then in the Latin classes.


Thinking back to my Latin classes at school, I doubt we ever read anywhere near as much as a book and that's over several years. I almost bought Familia Romana after watching the video just to see if it's as easy to get through as he makes it sound, but I don't want to distract myself :roll:
Last edited by Cenwalh on Wed May 08, 2019 7:59 pm, edited 1 time in total.
0 x
Double SC films: 200 / 200 (updated 2022-07-28)
Double SC books: 34 / 200 (updated 2022-07-28)

Cenwalh
Green Belt
Posts: 267
Joined: Thu Mar 28, 2019 9:14 am
Location: UK
Languages: English (N), Spanish (C1), Catalan (B2).
Language Log: https://forum.language-learners.org/vie ... 15&t=12467
x 848

Re: One million sentences to mastery?

Postby Cenwalh » Wed May 08, 2019 7:57 pm

Stefan wrote:I have never heard these numbers but for people interested, I would recommend the research done by Paul Nation. He's done countless of studies on vocabulary acquisition, extensive reading, word families and comprehension. He goes through the number of words students and kids know, how many times you need to come across a word to remember it, how many books you'd have to read, etc. He released a book in 2017 titled "How Vocabulary is Learned" but I haven't read it yet.

With this said, I don't know how well-founded his research is.


Would you believe that I linked to a Paul Nation study in my log today? I've read a few of his articles along with some other authors. His papers are very well cited and interesting, I would also recommend anyone check some out (or at least the abstracts).
0 x
Double SC films: 200 / 200 (updated 2022-07-28)
Double SC books: 34 / 200 (updated 2022-07-28)

themethod
White Belt
Posts: 40
Joined: Fri Apr 10, 2020 3:38 pm
Languages: English (N), Russian (Upper Intermediate), Spanish (Upper Intermediate)
x 124

Re: One million sentences to mastery?

Postby themethod » Sat Apr 25, 2020 12:41 pm

I came across this article by Antimoon, which also proposes the "one million sentences" idea.

How much input did I get? It took me about 3 years to get from basic English skills to fluency. During those 3 years, I was exposed to about 1,000,000 English sentences (not necessarily different sentences). About 400,000 of these were written sentences (books, SRS reviews, dictionaries, classroom reading); 600,000 were spoken sentences (TV, recordings, listening to teachers, listening to my American cousin, classroom listening).

Note that these are very rough estimates. The actual number of sentences that I got during that 3-year period may well have been 700,000 or 1,500,000.


http://www.antimoon.com/how/input-howmuch.htm

What I found particularly interesting is his math on how to hit that goal:

So if you want to follow in my footsteps, you’ll have to get about 60 pages of written English and 6 hours of spoken English per week — for three years. (I am assuming you already have some basic English skills that enable you to understand this article. If you are a total beginner, you will have to get to that level first.)


He lists 60 pages as 1,600 sentences and 6 hours of "non-stop" audio as 4,800 sentences.

Using those numbers, if you read 60 pages and listened to 1.5 hours of audio per day, you'd hit one million sentences in just under a year. That's roughly 21,900 pages and 548 hours of audio.
4 x

User avatar
Voytek
Green Belt
Posts: 407
Joined: Fri May 13, 2016 3:36 pm
Location: Chiang Rai (Thailand)
Languages: polski (N)
English(C2)
español(C2)
svenska (C1)
日本語 (A1)
ภาษาไทย (dabbling)
x 346

Re: One million sentences to mastery?

Postby Voytek » Sun Apr 26, 2020 11:29 am

Similar article on reading in Japanese with some interesting sources for readers:
http://www.kotobites.com/2017/06/21/tadoku-japanese-reading/
0 x
Exposure to Swedish-RL-building stage: 30 / 300
Exposure to Spanish-RL-final stage: 300 / 1500

User avatar
Iversen
Black Belt - 4th Dan
Posts: 4768
Joined: Sun Jul 19, 2015 7:36 pm
Location: Denmark
Languages: Monolingual travels in Danish, English, German, Dutch, Swedish, French, Portuguese, Spanish, Catalan, Italian, Romanian and (part time) Esperanto
Ahem, not yet: Norwegian, Afrikaans, Platt, Scots, Russian, Serbian, Bulgarian, Albanian, Greek, Latin, Irish, Indonesian and a few more...
Language Log: viewtopic.php?f=15&t=1027
x 14962

Re: One million sentences to mastery?

Postby Iversen » Mon Apr 27, 2020 10:16 am

Sentences is actually not a sensible unit of measure since they vary enormously in length. There are some really long and complicated sentences in the speeches of Cicero (and also in medieval books like Gesta Danorum by Saxo Grammaticus), whereas those in comics tend to be rather short. Better count words or books or shelf meters or time spent on reading (minus time spent on looking words up). In short: anything but sentences...
6 x

Cenwalh
Green Belt
Posts: 267
Joined: Thu Mar 28, 2019 9:14 am
Location: UK
Languages: English (N), Spanish (C1), Catalan (B2).
Language Log: https://forum.language-learners.org/vie ... 15&t=12467
x 848

Re: One million sentences to mastery?

Postby Cenwalh » Mon Apr 27, 2020 12:30 pm

Iversen wrote:Sentences is actually not a sensible unit of measure since they vary enormously in length. There are some really long and complicated sentences in the speeches of Cicero (and also in medieval books like Gesta Danorum by Saxo Grammaticus), whereas those in comics tend to be rather short. Better count words or books or shelf meters or time spent on reading (minus time spent on looking words up). In short: anything but sentences...


What is a sensible measure really depends on what you're measuring it for. I like sentences as a sentence is a whole message received/understood. You could say "I understood 100 messages today". The same can't really be said for words or books or time.

I also like sentences because as you get better, you can read or listen to more words per minute, but the sentences get longer, so it's like the number would stay relatively level. You could say time would be an equally good measure, but that ignores factors like a 90 minute film has no where near the amount of dialogue as 90 minutes of an audiobook.

You could pick on outliers of length as a reason why they're a bad measure, but a similar problem applies to all the other measures you stated: words have different densities in different media so it could take you ages to reach a goal someone else met in a short amount of time; time spent reading doesn't give you much information about the amount of input; books vary wildly in length and can have fluff like pictures.

On average with no biases in content choice, they are probably all a decent measure, but a million is a nice round number so I like it ;)
4 x
Double SC films: 200 / 200 (updated 2022-07-28)
Double SC books: 34 / 200 (updated 2022-07-28)

User avatar
ryanheise
Green Belt
Posts: 459
Joined: Tue Jun 04, 2019 3:13 pm
Location: Australia
Languages: English (N), Japanese (beginner)
x 1681
Contact:

Re: One million sentences to mastery?

Postby ryanheise » Mon Apr 27, 2020 2:28 pm

Cenwalh wrote:
Iversen wrote:Sentences is actually not a sensible unit of measure since they vary enormously in length.


On average with no biases in content choice, they are probably all a decent measure, but a million is a nice round number so I like it ;)


True, and true.

On the "small" scale, the variability in sentence length can be significant. This could be relevant if, say, a beginner (who can cope with fewer sentences per day) commits to studying 10 new sentences per day, but due to this variability, today's study session ends up with lots of short sentences and takes 10 minutes, while tomorrow's study session ends up with lots of long sentences and takes 2 hours and begins hurting your brain. So for daily consistency you may be better off counting words (or even counting "new" words) which increases the scale and makes the variance less significant.

Of course, once we're talking a MILLION sentences, surely things will average out ;-) . Sentences will have an average length in spite of this enormous variability, and this is also true for words. When we measure how many words per minute you can read or type, we measure it based on a standard average word length of 5, rather than counting how many words you actually read or typed.
2 x


Return to “General Language Discussion”

Who is online

Users browsing this forum: emk, tastyonions and 2 guests