Extensive reading and reading speed

[Dr Mack Rettosy]

Quote

I think 100cpm at sometime between 500k -> 1m is about right. 

 

I think this is probably a good estimate but depends on factors such as listening proficiency, reading style (intensive vs extensive), and chosen reading materials.

 

At ~600,000 characters my average reading speed is sitting at ~55 cpm. The linear regression of my data set is admittedly weak (R^2=0.397) but suggests by 1m characters a reading speed of ~92cpm. FWIW, I'm lazy. I'm very much a point and click reader with a high tolerance for ambiguity. I can't even be bothered to make flashcards. On the other hand, an intensive reader would probably be getting more from their material and probably improve faster, at least when cumulative characters is being used as the independent variable.

 

 

[alantin]

On 1/22/2022 at 10:01 PM, Dr Mack Rettosy said:

Quote

I think 100cpm at sometime between 500k -> 1m is about right. 

 

I think this is probably a good estimate but depends on factors such as listening proficiency, reading style (intensive vs extensive), and chosen reading materials.

 

 

Seems optimistic to me if you are aiming to have 100cpm speed with native materials. I'm averaging 80-90 cpm with WoT and can get about 110 cpm with a graded reader. I also think my reading style and tolerance for ambiguity is probably very similar to @Dr Mack Rettosy's. My own chart seems to complement yours quite nicely! 

 

I have only gathered data for about three or four months during which time I've read about 500k characters. When I began, my speed was between 60-70 cpm and I had read a couple of books already so I had done something between 1M and 2M characters already. So my chart probably shows my speed between 2M and 2.5M characters. My forecasts are promising 100cpm with the material I'm currently reading in about another 500k characters so at around 3M characters.

[Dr Mack Rettosy]

@alantin I need to figure out how to quote on these forums lol.

 

Quote

Seems optimistic to me if you are aiming to have 100cpm speed

 

I made an exponential with a better fit (R^2=0.68) which moved the number higher, just north of 3.0 M characters. Really neat to see all our data being consistent, even with Pinion's OP, with his reading speed being ~75cpm at 1.5-2.0 M cumulative characters!

[phills]

On 1/23/2022 at 6:47 AM, alantin said:

When I began, my speed was between 60-70 cpm and I had read a couple of books already so I had done something between 1M and 2M characters already.

 

I love seeing all of this data shared.  I think it'll really help us piece together the whole trajectory of the learning-to-read experience.  Of course individual experiences will vary but it's nice to see the range.

 

When I started reading, I had read a bunch of stuff already, so I wasn't starting fresh   I knew ~2000-2500 characters before I started my extensive reading campaign.  I didn't do many graded readers (although I did a few), but I read a bunch of movie/tv subtitles + Chinese language comics.  Just not native materials in volume. 

 

@alantinA couple of books is probably only about 150-300k chars.  I estimate old school children books (pre Harry Potter bloat) are about 50-100k chars, while medium length "novels" are 100-300k. 

 

If you haven't have started reading in volume, 1M chars is a lot because you tire out quickly.  So you're reading slowly + you pause/rest a lot.  You repeat, you spend time looking stuff up.  At least as I recall my experience, the first million chars subjectively took a long time. 

 

E.g. Similarly, I feel like I've done more focused listening now this month than I've ever done, but I think it's still only a few hundred thousand characters.  For listening too, there's an equivalent.  There's listening time -- and then there's the zoning-out time, as you exhaust your attention span, even if you refuse to pause.  Then there's the occasional repeat / look stuff up.  And just like for reading, I've already listened to a lot of stuff before this extensive listening campaign.

 

@Dr Mack Rettosy Are you reading shorter material?  Are you switching a lot between different types of content?  As you get used to a book, an author, a genre, it'll become easier for you and you'll speed up.  You can easily get 30%+ improvement over a course of a book, from the beginning to the end.

 

E.g. I still read "news" much slower than I read "books" because I have a long warm-up period. By the time I hit my stride, the news article is over.  Same, but less dramatically so with short stories. I often don't feel comfortable in a book until 50k chars in, before I get up to my cruising speed.

[Publius]

On 1/23/2022 at 8:54 AM, Dr Mack Rettosy said:

I need to figure out how to quote on these forums lol.

Haha, easy. You select text using either mouse or finger (click and hold -> drag -> release, or longpress -> adjust scope), a "Quote selection" box will pop up, click on it and it will send you to the reply field along with your quote.

 

[realmayo]

I'm puzzled how this data is matching up. Pinion - if I remember right - already spoke Chinese, he/she just didn't read. I can't see how that experience is super-comparable to most of us who didn't grow up speaking the language. Similarly, you'd expect that people who pick relatively easy level texts will, over time, pick more difficult ones, so like isn't being compared with like. Also, you'd expect that someone who spent a lot of time watching Chinese TV would end up with a faster reading speed, as well as better listening comprehension, than someone who didn't!

 

I do think collecting data about one's own progress is helpful, for motivation and self-testing/improving, as well as getting a sense of how easy/difficult certain writers are, and of course for just figuring out how long it'll take to read a certain book. Also it's encouraging to see other people sharing data here beacuse it's always encouraging to see others' progress. I am just cautious about people basing their longer-term study plans ('I need to spend x amount of time reading to reach this level') by extrapolating from this data.

 

Simple example: if you spent 10 minutes every day practising on improving your reading speed, then your reading speed will probably increase faster than if you spend 100% of your reading time just reading in a normal way.

[alantin]

On 1/23/2022 at 7:12 AM, phills said:

@alantinA couple of books is probably only about 150-300k chars.  I estimate old school children books (pre Harry Potter bloat) are about 50-100k chars, while medium length "novels" are 100-300k. 

 

I did some proper digging on this. I read parts of the first Harry Potter, 流浪地球, and 三体1 on Lingq in the first half on 2020. Before that I had read and reread a few graded readers that were probably around 10k characters. Then I switched over from lingq to reading in a word document while playing with the fonts while reading to add pinyin above unknown characters. I read the 6th WoT book that way beginning in December 2020. I read the first half twice (following the actual paper back copy) and a small part of the beginning of the second half. I did that for about half a year or so and stopped. Then I picked up again in October 2021 using CTA this time and that's were my current data collection begins.

 

Lingq shows me that I've read 162 888 words there so maybe I arrive at characters by multiplying that by 1.5 so 244 000 characters. I don't have an exact character count for the 6th WoT book but they range between 500k and 600k characters.

 

So it seems 2M is way too high, but adding together those above and considering some random stuff I've read online too, now I'm quite certain I was at about 1M characters when I began collecting my data in last October. I also have one reading speed figure in my notes from 6 January 2021 so almost exactly a year ago. I did 63 characters per minute reading the chapter three of the 6th WoT book. But those days I included punctuation in the character count and I understand CTA leaves them out, so to make it comparable, it's probably about 10%-15% lower so around 55cpm

 

 

On 1/23/2022 at 7:12 AM, phills said:

If you haven't have started reading in volume, 1M chars is a lot because you tire out quickly.  So you're reading slowly + you pause/rest a lot.  You repeat, you spend time looking stuff up.  At least as I recall my experience, the first million chars subjectively took a long time. 

 

I agree. This is pretty much my experience too.

 

 

On 1/23/2022 at 9:09 AM, realmayo said:

I'm puzzled how this data is matching up. Pinion - if I remember right - already spoke Chinese, he/she just didn't read. I can't see how that experience is super-comparable to most of us who didn't grow up speaking the language. ......

 

I am just cautious about people basing their longer-term study plans ('I need to spend x amount of time reading to reach this level') by extrapolating from this data.

 

It is interesting how other people's progress seems to be following Pinion's. To me this tell that reading speed is not that much a function of vocabulary size but of ability to parse the characters quickly, which is an unrelated skill. I bet native Chinese school kids aren't much faster than we are either after reading their first million characters even if comprehension level was higher.

 

In any case, a few people sharing their own data is hardly statistically trustworthy sample, but it does offer insight and in the very least tells you that you do need to dedicate a LOT of time to reading to get good at it. It also does give you motivation when you're not feeling progress. My gradient may not be exactly the same as someone else's, but the graphs look very similar.

[Jan Finster]

Forgive me if it is mentioned somewhere in this thread: how do you measure your reading speed?

 

a) Do you time yourself over 1 minute? 

b) Do you read 1000 characters and stop the time?

c) Do you run a timer whenever you read a book and average characters per book divided by total time?

d) xyz?

 

My reading speed for 100 characters or one minute may not mean much as I will not be able to keep that pace over the course of a book. So, time per book (option "c") sounds most meaningful to me. A bit like marathon time vs 100 meter sprint time....

 

[alantin]

On 1/23/2022 at 10:25 AM, Jan Finster said:

c) Do you run a timer whenever you read a book and average characters per book divided by total time?

 

I think most people posting data on this thread are doing C. Yours truly at least is.

 

I did those one minute spot reading speed tests at some point. They don't mean anything. I'd try to read faster than I actually can to get a better score, and that's not useful data.. All my results are averages of reading one full chapter and in reading time can mean anything between half an hour and two or three hours. There are some darn long chapters in WoT!

 

[alantin]

@Jan Finster, I checked if reading time correlates somehow with reading speed. Read times being anything between half an hour and three hours, the correlation for me is quite weak (-0.27). It suggests that I read shorter chapters at a little higher speed, but the relationship between reading speed and read time of a chapter is far from consistent.

 

* 1.0 is perfect correlation, 0 is no correlation at all, and -1.0 is perfect negative correlation (as one variable increases, the other one decreases).

[phills]

On 1/23/2022 at 4:25 PM, Jan Finster said:

c) Do you run a timer whenever you read a book and average characters per book divided by total time?

 

Yep!  I have a script that does that; like a chess clock.  It also counts punctuation because it just counts characters in a file.  Some of the people here don't count punctuation, some do, so you can argue our counts are not exactly comprable.  But that only affects 10-15% of the result.  It wouldn't affect the order of magnitude, or the relative speed-up.

 

On 1/23/2022 at 3:09 PM, realmayo said:

Simple example: if you spent 10 minutes every day practising on improving your reading speed, then your reading speed will probably increase faster than if you spend 100% of your reading time just reading in a normal way.

 

This I'm not sure I believe.  I think the speed reading exercises only help if your understanding is outpacing your reading; maybe because you're hesitant or perfectionist about pronunciation or you're stuck in a habit.  Then the exercise might jog you out of your routine.

 

I feel like I'm already reading as fast as I can understand stuff.  If I force myself to read faster, I'll just understand less.  I feel like I can increase speed only slowly by learning to process the language faster in my head, and getting better at the language as a whole.  My smoothest reading experiences are when the text draws me in; while my most clumsiest experiences are when I try to overpower the characters. 

 

On 1/23/2022 at 4:19 PM, alantin said:

To me this tell that reading speed is not that much a function of vocabulary size but of ability to parse the characters quickly, which is an unrelated skill. I bet native Chinese school kids aren't much faster than we are either after reading their first million characters even if comprehension level was higher.

 

This is closer to what I believe, although from my half-a$$ed internet research, kids are just better at learning to go fast, than adults are (maybe 50%-100% better).  If you first learn to read as an adult, you lose most of that 50% kid boost, even if you can speak the language.  While if you read a lot in a second language as a kid (millions of words), you're probably a decent reader, even compared to natives.

 

What I'm curious is how fast people like Arnold Schwarzeneggar or Henry Kissinger read.  They never lost their accents so they learned English at too late an age for that, but did they learn reading at an early enough age to gain high-end native speeds?  I imagine at least Kissinger reads really fast, but who knows?  Wikipedia says Kissinger came to the US at 15, and Schwarzeneggar 21. 

 

[Moshen]

Quote

What I'm curious is how fast people like Arnold Schwarzeneggar or Henry Kissinger read.  They never lost their accents so they learned English at too late an age for that, but did they learn reading at an early enough age to gain high-end native speeds?  I imagine at least Kissinger reads really fast, but who knows?  Wikipedia says Kissinger came to the US at 15, and Schwarzeneggar 21. 

My husband came to the US at age 33.  English was his third language, and he mastered it well enough to get two masters degrees, and work a high-level tech job in English.  The only trouble he has understanding the radio/TV/newspaper news or ordinary conversation is when there's an unusual word or cultural reference.  But when there's something practical he needs to skim for quick comprehension, to get the overall gist, he asks me to help.  So it seems like he doesn't read fast enough for that kind of task.

[Jan Finster]

On 1/23/2022 at 1:48 PM, phills said:

  I have a script that does that....

 

On 1/23/2022 at 9:50 AM, alantin said:

I checked if reading time correlates somehow with reading speed

 

On 1/23/2022 at 1:54 AM, Dr Mack Rettosy said:

I made an exponential with a better fit (R^2=0.68).....

 

You guys really know how to make others feel inferior... ?

[Dr Mack Rettosy]

On 1/22/2022 at 11:12 PM, phills said:

Are you reading shorter material?  Are you switching a lot between different types of content?

 

I started with Rainbow Bridge graded readers, then Mandarin Companion graded readers, now a mix of translated materials (AA Milne and Roald Dahl) and native materials (孙幼军, 夏正正, and now the works of 曹文轩).

 

On 1/23/2022 at 12:25 AM, Publius said:

You select text using either mouse or finger

 

Thank you! ?

 

On 1/23/2022 at 1:09 AM, realmayo said:

I'm puzzled how this data is matching up. Pinion - if I remember right - already spoke Chinese, he/she just didn't read. I can't see how that experience is super-comparable to most of us who didn't grow up speaking the language. Similarly, you'd expect that people who pick relatively easy level texts will, over time, pick more difficult ones, so like isn't being compared with like. Also, you'd expect that someone who spent a lot of time watching Chinese TV would end up with a faster reading speed, as well as better listening comprehension, than someone who didn't!

 

I do think collecting data about one's own progress is helpful, for motivation and self-testing/improving, as well as getting a sense of how easy/difficult certain writers are, and of course for just figuring out how long it'll take to read a certain book. Also it's encouraging to see other people sharing data here beacuse it's always encouraging to see others' progress. I am just cautious about people basing their longer-term study plans ('I need to spend x amount of time reading to reach this level') by extrapolating from this data.

 

I second this caution. Progress will be different for everyone depending on their background and study habits. For myself, the "zero" character point is not when I started learning Chinese. Looking at my log, I already had seven months of study with 600+ hour which included a lot of immersive listening and reading in pinyin.

 

On 1/23/2022 at 2:25 AM, Jan Finster said:

Forgive me if it is mentioned somewhere in this thread: how do you measure your reading speed?

 

a) Do you time yourself over 1 minute? 

b) Do you read 1000 characters and stop the time?

c) Do you run a timer whenever you read a book and average characters per book divided by total time?

d) xyz?

 

You guys really know how to make others feel inferior... ?

 

Hey friend, you are giving us (or at least me) way too much credit. If you're engaging with Chinese material every day, that's all that matters.

 

Anyway to answer your specific question, I pretty much do C). Keep in mind, I read using an iPad and the Pleco document reader. Here is my very simple workflow when I start a new book:

 

Choose my new book (always fun).

Open the book's ePub on Calibre.

Open this character counter. I prefer this one because it doesn't count punctuation.

Open my excel spreadsheet.

Copy paste each chapter's script into the character counter then into my excel spreadsheet, then sum to get total character count. This takes about 2 minutes to do the entire book.

 

On my iPad notes, write down the date I start the book.

Read book on Pleco,

On my iPad notes, write down the date I finish the book.

Go to Settings > Screen Time > See All Activity. The amount of time I spent on the Pleco app is given for each week or each day. Add it up to get total time (in minutes) spent reading the book.

Take the total characters divided by total minutes = characters per minute.

 

This takes all of less than ten minutes per book. And echoing alantin, I find it enjoyable and motivating. Hope that helps!

 

 

[realmayo]

On 1/23/2022 at 12:48 PM, phills said:

I think the speed reading exercises only help if your understanding is outpacing your reading

 

I think there are two principles at work here.

 

1. If you want to get fluent at an isolated skill, use easy materials.

2. If you want to improve your all-round language ability, use slightly challenging materials.

 

1. To improve the isolated skill of reading-quickly, spend a bit of time every day trying to read easy stuff quickly. You should soon find that you will start reading normal materials more quickly too.

 

2. To improve your understanding of the Chinese language (vocabulary, syntax, nuance, culture etc) read slightly challenging materials.

I don't think you can do 1 and 2 by using the same kind of texts.

 

On 1/23/2022 at 8:19 AM, alantin said:

and in the very least tells you that you do need to dedicate a LOT of time to reading to get good at it.

 

I think the data tells you how much time you need to dedicate to reading if all you want to do is improve your reading speed but not improve your language ability.

 

What the data hides from you is that you'll make much more rapid improvement in reading if, in addition to just reading, you spend time studying the Chinese language too!

 

I think that's logical, right? Otherwise teachers would just assign reading texts and leave their students alone. My own experience is that reading ability (including of course reading speed) improves the fastest after prolonged and intensive language study (reading/listening/speaking/writing).

 

That's why I'm cautious about drawing conclusions from this data.

 

Also from a statistical point of view, there are way too many confounding factors. For instance, when it comes to selecting difficulty of reading materials, someone who is rapidly improving their language abilities will naturally select increasingly difficult texts; someone making almost zero improvement will not select increasingly more difficult texts. But the graphs could well show reading speed increasing at the same speed for both students.

[alantin]

On 1/23/2022 at 4:23 PM, Dr Mack Rettosy said:

On 1/23/2022 at 10:25 AM, Jan Finster said:

You guys really know how to make others feel inferior... ?

 

Hey friend, you are giving us (or at least me) way too much credit. If you're engaging with Chinese material every day, that's all that matters.

 

LoL. It's just the Chinese Forums nerd squad gathering. ?

 

My workflow is pretty much the same as @Dr Mack Rettosy's but I collect the stats from CTA to my excel sheet at the beginning of a chapter and fill a bit more after finishing it. It makes it possible to throw different kinds of data analysis at it with built in excel tools. Also I don't trust the time spent on app stats from phones. I use Toggl app on iphone to start the timer when I start to read and stop it when I stop and record the final time spent when I finish a chapter.

[alantin]

On 1/23/2022 at 5:06 PM, realmayo said:

What the data hides from you is that you'll make much more rapid improvement in reading if, in addition to just reading, you spend time studying the Chinese language too!

 

I think that's logical, right? Otherwise teachers would just assign reading texts and leave their students alone. My own experience is that reading ability (including of course reading speed) improves the fastest after prolonged and intensive language study (reading/listening/speaking/writing).

 

That's why I'm cautious about drawing conclusions from this data.

 

We are specifically discussing "Extensive Reading and Reading Speed" in this thread.

 

I don't think anyone has suggested reading speed and language ability to be synonymous and I haven't seen anyone write about exclusively concentrating on improving reading speed and expecting that to improve overall language ability... Though I personally believe reading to to be the single best activity for building the foundation for improvement in the other areas of the language and for that greater reading speed provides you with greater amount of input in a given space of time, so higher reading speed provides benefits over lower reading speed. As a variable it is also a convenient proxy to gauge the improvement of your reading skills due to it being very easy to record, though it is only one small part of the skill.

 

In my case reading takes about 27% of all my Chinese language activities (listening: 43%, speaking: 11%, handwriting: 12%, typing: 7%), so I'm hardly doing any kind of exclusive reading training. Also of-course all of the skills mutually affect each other in some way and I personally find handwriting to be especially helpful with character recognition, but I'm uncertain about how to record the quality of those activities so I could see their effect on reading speed specifically... I am currently recording the amount of time I spend on each, so maybe I could add that to my reading progress data and see if there are correlations... I'm skeptical I would find any though. So far the only variables I've seen correlating with reading speed in any significant manner are the "cumulative characters read" and the "difficulty of the material".


 

On 1/23/2022 at 5:06 PM, realmayo said:

1. To improve the isolated skill of reading-quickly, spend a bit of time every day trying to read easy stuff quickly. You should soon find that you will start reading normal materials more quickly too.

 

I think this might work. I did a test reading an easy graded reader some time ago as quickly as I could. My speed was a lot faster than my normal speed and the speed gain DID carry over to my normal reading, but the effect seems to have worn off after that so a one off isn't going to make wonders to your overall reading speed. Doing it every day should. But my only problem is that I don't find those graded readers very interesting any more...

[realmayo]

On 1/23/2022 at 4:06 PM, alantin said:

We are specifically discussing "Extensive Reading and Reading Speed" in this thread.

I suppose I'm suggesting that you're unlikely to get an accurate sight of how your reading speed improves relative to the cumulative number of characters you have read, because the material you're reading should increase in difficulty over time - and then, at an advanced level, it will kind of stop getting more difficult (unless you force yourself to read technical manuals or philosophical treatises).

 

I wasn't suggesting you weren't doing anything else apart from reading. But if you're only measuring reading, then either all the other studying you're doing is having zero effect on your reading (highly unlikely) or your reading speed is correlated to a lot more than the cumulative number of characters you've read - your x axis is being affected by a lot more than your y axis!

 

I'm not saying it's not worthwhile looking back at data. I'm just specifically sceptical that you should be taking future-looking extrapolations too seriously (i.e. how much you have to read to get to a certain reading ability).

 

 

On 1/23/2022 at 4:06 PM, alantin said:

Doing it every day should. But my only problem is that I don't find those graded readers very interesting any more...

I think the idea is that you re-read the same short text until you're at the desired speed. Then move onto the next text. Very very boring!

 

[realmayo]

How do people find character-uncertainty slows them down? If there are certain characters that you sometimes have to pause over, do you make time to learn/relearn them, or do you just expect that by more and more reading you'll end up solidifying their meaning/pronunciation?

[phills]

On 1/26/2022 at 5:56 PM, realmayo said:

How do people find character-uncertainty slows them down? If there are certain characters that you sometimes have to pause over, do you make time to learn/relearn them, or do you just expect that by more and more reading you'll end up solidifying their meaning/pronunciation?

 

It does for me.  Over time, my memory degrades and I start recognize characters partly by context.  Particularly characters with similar looking / sounding components, or you see together a lot. 

 

Reading starts to subjectively feel like it takes a lot of mental "energy" as I have to puzzle things out.  I read slower, until I warm-up. 

 

When it gets too bad, I go back to drilling characters and reading all of a sudden reading feels like it energizes rather than drains energy.  Characters come easy and the text pulls me in, rather than feeling like I have to overpower it.

 

Doing more and more reading does solidify meaning, but it's doesn't help with confusing characters.  When I do re-drilling, my mistakes are usually of this type, rather than drawing a pure blank.  Similar looking 暖 vs 暧, similar sounding 漫 vs. 慢 or two parts of a word 继 vs 续.