Extensive reading and reading speed

[phills]

I just finished 异兽志.  It's my first case of measuring intensive reading rather than extensive reading.  How does this affect reading speed?

 

beast1: 51.64 mins, 8918 chars, 172.70 cpm; (re-read) not measured
beast2-1: 56.19 mins, 9185 chars, 163.46 cpm; (re-read) 49.19 mins, 180.87 cpm
beast3: 63.92 mins, 9931 chars, 155.37 cpm
beast4: 54.11 mins, 9101 chars, 168.19 cpm
beast5: 62.17 mins, 8697 chars, 139.89 cpm
beast6-1: 64.70 mins, 9300 chars, 143.74 cpm; (re-read) 75.43 mins, 126.51 cpm
beast7: 81.67 mins, 12642 chars, 154.79 cpm
beast8: 76.70 mins, 11788 chars, 153.69 cpm
beast9: 97.83 mins, 15681 chars, 160.29 cpm
Total (first time): 10.15 hrs, 95,243 chars, 156.41 cpm

 

I had to re-read 3 of the 9 chapters.  I started out at my normalish 180cpm and found I couldn't understand what I was reading, and had to read each chapter twice.  So I decided to slow down from that point on, instead of speeding up.  Chapter 6 was particularly that way, I actually read it slower the second time than my first -- fully committing to the Go Slow approach.

 

I decided I couldn't read in 3rd gear (180cpm) with this material.  I could only understand it if I read in 1st gear (120) or 2nd gear (150).  Plus I needed the forum thread to fully understand the material.

 

Why was the book hard?

 

1. The language in the book is on the hard side.  I gathered 156 vocab words / chengyu from it, or 1 in ~600 chars.  For my normal extensive reading, I'm gathering 1 per 1000-2000 chars.  As @alantin notes in his correlation, the more words you don't know, the slower you read. 

 

(This isn't all the words I don't know / have to guess -- just those I gather, which is a subset.)

 

2. It uses compact language and is full of elegant phrasing.  The explanatory sections were often in 文言文 (the very sections that explain how every thing fits together).  Plus there was information hiding with pronouns, words with double meanings etc.

 

3. The plot itself is complicated and full of (fun) twists.  It would hard to understand, and would require backtracking, even in English, if you want to see how all the details fit together. 

 

The first reviewer on weread (presumably a native speaker) thought it was confusing too!

 

 

https://weread.qq.com/web/reader/4d832910718245534d810eb

 

Reading something hard does hurt your confidence.  Halfway through, I also went back to drilling characters, because I was frustrated that the reading wasn't coming easy to me, and requiring too much brain power.  迷迷糊糊! 

 

[黄有光]

On 11/27/2021 at 2:26 PM, phills said:

The explanatory sections were often in 文言文 (the very sections that explain how every thing fits together).  Plus there was information hiding with pronouns, words with double meanings etc.

I've been told by a highly advanced L2 Mandarin speaker on Reddit that one develops a general understanding of the fundamentals of 文言文 as a matter of consequence of learning Mandarin to a high level. In your opinion, is such incidental knowledge sufficient to understand these sections, or is more formal study required?

[phills]

On 11/29/2021 at 8:49 AM, 黄有光 said:

In your opinion, is such incidental knowledge sufficient to understand these sections, or is more formal study required?

 

Well I never studied it, and the passages probably arent fully 文言文.  I could guess a lot of the meaning, but there were some tough sections.  It's like reading Shakespeare and seeing a bunch of thou's, such's and mayest's all of a sudden. 

 

If you understand your single characters well (I suppose those are called morphemes), especially the multiple meanings / usages each single character can have, you can decipher a lot of it.  I'm not an expert on what exactly 文言文 is, but to me, it's much more compact and uses many single character words rather than double character words.  Thus my taking a short detour to do single character drills again while reading it.

 

The forum thread has a lot of explanations, by chapter, in case you get stuck.  They're all gated behind spoiler tags for the most part, so it won't spoil you to check it after each chapter. 

[Publius]

Well, to get a feel of 文言文: ?

 

《聊齋志異·狐聯》

　　焦生，章丘石虹先生之叔弟也，讀書園中，宵分有二美人來，顏色雙絕。一可十七八，一約十四五，撫几展笑。焦知其狐，正色拒之。長者曰：「君髯如戟，何無丈夫氣？」焦曰：「僕生平不敢二色。」女笑曰：「迂哉！子尚守腐局耶？下元鬼神，凡事皆以黑爲白，况牀笫間瑣事乎？」焦又咄之。女知不可動，乃云：「君名下士，妾有一聯，請爲屬對。能對我自去：戊戍同體，腹中止欠一點。」焦凝思不就。女笑曰：「名士固如此乎？我代對之可矣：己巳連踪，足下何不雙挑？」一笑而去。

[phills]

I just finished 雪山飞狐.  It started out a real slog, but it ended up being quite readable.  I take back what I said about it being harder than 异兽志.  The stats:

 

snowfox1: 3.13 hrs, 25742 chars, 137.10 cpm
snowfox2: 3.71 hrs, 32791 chars, 147.29 cpm
snowfox3: 2.27 hrs, 22754 chars, 167.32 cpm
snowfox4: 2.35 hrs, 25593 chars, 181.37 cpm
snowfox5: 2.43 hrs, 24987 chars, 171.53 cpm
Total: 13.89 hrs, 131,867 chars, 158.27 cpm

 

1. About 42k characters in (in the middle of the snowfox2), suddenly everything clicked.  snowfox2 was going at about 120cpm (slower than snowfox1), and then -- all of a sudden -- I read the rest of it at 170-180 cpm (my typical 3rd gear speed).  After that, I read the story mostly in 3rd gear.

 

Some of it is because the story started having more dialog rather than pure description & scene setting.  But before that point, even the dialog was slow.  After that, even the descriptions went by faster.

 

2. I'm struck by how similar the pattern is with 三体 3.  It took me about 50k characters there to get my sea legs, and then it became smooth sailing.

 

雪山飞狐 is still more difficult than 三体 3.  The smooth sailing afterwards is still not quite as smooth, and I collected more vocab from it than 三体 3.  I ended up collecting 128 out of 130k characters, so 1 entry for every 1000 chars.  That's my high end for "extensive reading". 

 

3. The 2 main factors that make 雪山飞狐 hard are:

 

a. The vast cast of characters.  Once I figured out who everyone was, and which cliques they belonged to, and who hated who, it became much easier to understand.  Surprisingly, the wiki isn't as great for 雪山飞狐 as for other 金庸 novels.

 

b. The literary / kung-fu jargon.  I'm sure if I read more of it, it'll become good extensive reading material.  I'm not quite there yet, but it's not slog / study material that I originally feared either.

 

I'll post some of my notes in a book thread so others can get a head start. 

[phills]

I just finished 鸳鸯刀, a 金庸 short story (really, only 35k chars!).  I was curious if my newly acquired 金庸 speed would be transferrable...  And it was! 

 

It seems there really is sort of a hump, you have to get over, to acclimatize yourself to his writing style & jargon.  

 

yuanyang1: 105.41 mins, 18510 chars, 175.60 cpm
yuanyang2: 90.18 mins, 16846 chars, 186.80 cpm
Total: 3.26 hrs, 35,356 chars, 180.77 cpm

 

Of course, the comprehension was much better as well.  I didn't spend the first third confused as to what was going on. 

 

I collected 17 vocab entries out of 35k chars, so roughly 1 out of 2000.  I encountered more unknown words than that, describing weird yells, facial expressions, kungfu moves, animals/foods, but I could quickly guess that's what those were and skipped over them.

[rickardg]

I just finished the first third of 平凡的世界, but my reading speed was on average an abysmal 80 cpm (max: 95 cpm, min: 60 cpm, std.dev.: 9). I really need to improve that...

 

Also when crunching the numbers I discovered that my txt-file (obtained from the underbelly of the internet) only has about 822,000 characters while Baidu claims it should have about 1,040,000. I might have to get a paper copy to make sure I'm not missing some chapters.

 

 

 

 

[alantin]

I'm 356,642 characters into the Wheel of Time after beginning it from the start on 18 October and have reached the average abysmal reading speed of 89 cpm (max: 96 cpm, min: 68 cpm, std.dev.: 6,9)

 

I did a little bit of a test to see how my reading speed compares when reading something easy and read the easy readers version of the Journey to the Center of the Earth. The difference was quite drastic. My average speed ready WoT over the previous 10 chapters was 79 cpm, but I read the graded reader at 112 cpm. I also saw similar reading speed increase as reported by @phills. I read the next chapter of WoT at 93 cpm and over the next six chapters my average speed has been 90 cpm.

 

 

I have also asked a few native speakers to time their reading for me and got four responses so far. They have all ranged around between 500 and 600 cpm. One of my regular teachers, who reads a lot, timed herself with two different texts and averaged 520 cpm. She was really kind and asked two friends to do the same. One did 502 on a random text and another one, who my teacher described as a really fast reader, did 600 cpm. We later had an interesting discussion about reading speed and I also asked her if she subvocalizes during reading. She said that she does and she considered around 500 cpm to be a normal reading speed.

The tutor, I talked to the other day, said that she hadn't read a book in years and suspected her speed to be abysmal. I pasted 1000 characters of WoT for her from the chapter I was at, and she did 531 cpm.

 

Based on these discussions, I've made 100 cpm my first short term goal and 250 cpm my long term goal. I think half the native speed would be a really good accomplishment and I think it also would allow me to complete reading all questions on the HSK6 test without problems, should I ever take it.

If my forecast, based on my performance so far, is accurate I should reach 100 cpm by the end of the first book (around 500,000 chars) or latest by half way to the second book (775,000 chars). The forecast for 250 cpm at the moment is at about 3,000,000 chars with the 95% Lower Bound at 2,250,000 chars the 95% Upper Bound at 4,800,000 chars.

Edit: I took a look at older posts in this thread and I think I'm most likely going to be looking at the upper end of that range if not far beyond it. I'm looking forward to seeing how the numbers have updated after finishing the first two or three books by summer.

[phills]

Great data, @alantin.  I'm glad to see that reading something easier boosted your speed when you went back to WoT.  It's funny how that goes sometimes, but I totally felt that happen to me too.  (Even reading something in English fast, helped me one time, when I went back to Chinese, reading in the same environment).  

 

I think 100cpm at sometime between 500k -> 1m is about right.  250cpm though is hard --> I'm at 5million and I haven't hit it yet.  Even with easy stuff I'm at low 200s.  I'm targeting 8m to reassess, based on pinion's graph...

 

Reading fast as a second language learner is just HARD.  I think it might the second hardest thing after perfecting accents, which most people can't do even after a lifetime of speaking a second language.

 

250cpm is a worthy goal.  It's half native, but it's also natural speaking speed.  If you want to read the fastest subtitles without any problems, you need to get to 250cpm.  If you want to read faster than an audio-book, you need 250cpm.  If you want to converse about more complicated subject matter at normal speeds, you need to be able to process Chinese characters at 250cpm in your head.

[rickardg]

On 12/11/2021 at 11:25 PM, phills said:

Reading fast as a second language learner is just HARD.  I think it might the second hardest thing after perfecting accents, which most people can't do even after a lifetime of speaking a second language.

 

Yes, but I also blame writing system ?. I taught myself Italian a couple of years ago and while my reading speed probably isn't anywhere near native, I don't really feel it is holding me back as it does with Chinese. A bit depressing considering I've studied Chinese on and of for years (admittedly mostly off the last few years).

[alantin]

On 12/12/2021 at 12:25 AM, phills said:

250cpm though is hard --> I'm at 5million and I haven't hit it yet.  Even with easy stuff I'm at low 200s.  I'm targeting 8m to reassess, based on pinion's graph...

 

Clearly there is a curve of diminishing returns that my trend curve can't show yet. The 95% confidence interval should express it by widening the range of expected outcomes too, but data over a couple of hundred thousand characters clearly isn't enough to show it either yet millions of characters ahead, if the improvement is steeper in the beginning before it starts to flatten out. The graded readers story itself and the the boost from it are also prone to distort the forecast before the effects starts to even out. But I think I'm going to leave it there as an outlier since it was an interesting data point for my correlation calculations too.

[rickardg]

On 12/11/2021 at 11:00 PM, alantin said:

I did a little bit of a test to see how my reading speed compares when reading something easy and read the easy readers version of the Journey to the Center of the Earth. The difference was quite drastic. My average speed ready WoT over the previous 10 chapters was 79 cpm, but read the graded reader at 112 cpm. I also saw similar reading speed increase as reported by @phills. Then I read the next chapter of WoT at 93 cpm and over the next six chapters my average speed has been 90 cpm.

 

This is very interesting. I've never gotten into graded readers because, frankly, they are boring, but if the effects are this marked I'll give it a try.

 

I must also try to get my hands on some books for, say, middle school students.

[alantin]

On 12/12/2021 at 1:05 AM, rickardg said:

This is very interesting. I've never gotten into graded readers because, frankly, they are boring, but if the effects are this marked I'll give it a try.

 

I found them helpful to get going and read a few stories before jumping into the native books. Now I find them useful for gauging my progress, and reading that one story really gave me a confidence boost in my own reading ability. It was about the same length as a regular chapter in WoT but it was extremely easy to read in comparison as there truly was no single new character for me and the meanings of the few words I didn't know already were immediately apparent from their characters and the context. But I can't really see myself using them for anything else than for measuring my top speed every now and again exactly because I usually don't find them interesting enough.

[phills]

Ah Visible Progress!  After reading 2 tougher works, I went back to something within my comfort zone, 圆月弯刀 by 古龙. I exceeded 250cpm for the first time on a book!

 

curvedsaber1: 2.07 hrs, 28849 chars, 232.78 cpm
curvedsaber2: 2.03 hrs, 30228 chars, 248.28 cpm
curvedsaber3: 110.84 mins, 31156 chars, 281.09 cpm
curvedsaber4: 108.51 mins, 29160 chars, 268.73 cpm
curvedsaber5: 2.53 hrs, 35710 chars, 235.45 cpm
curvedsaber6: 114.03 mins, 27627 chars, 242.28 cpm
curvedsaber7: 2.21 hrs, 33383 chars, 251.68 cpm
curvedsaber8: 2.01 hrs, 32133 chars, 266.53 cpm
curvedsaber9: 2.23 hrs, 34115 chars, 255.35 cpm
curvedsaber10: 104.11 mins, 26562 chars, 255.13 cpm
curvedsaber11: 48.53 mins, 12767 chars, 263.07 cpm
Total: 21.17 hrs, 321,690 chars, 253.27 cpm

 

This brings my cumulative character count to 5.46million.  It was a real stretch goal to hit 250cpm before the end of the year (I targeted 200+ by 5m), so I'm ecstatic about this milestone.

1. The experience was very similar to reading 兄弟2 a couple of months ago.  For the most part, I was still reading in 4th gear -- just 30 cpm faster than last time.  I was still occasionally tripping up on characters, making recognition mistakes,  looking up the random word or 2.  It's only when I ended the session and saw the time that I realize that I've gone significantly faster than before.

2. Just like 兄弟2, there was one spot in the book when I hit 5th gear.  Halfway through curvedsaber3, I was reading at 300cpm.  But just as soon as I became aware that passages were zooming by, it was gone.  I never got back there afterwards.

3. Comparing the difficulty of 古龙 with 金庸, one subjective answer is 古龙 is 金庸 is 40% harder than 古龙.  Even after I  got comfortable reading 金庸, I read it at 180cpm / gear 3, while I read 古龙 at 250cpm / gear 4 to 5.  About 40% faster.

 

Vocab-wise, I collected 100 vocab words from 300k chars, so a very comfortable 1 in ~3000.

 

4. Now to see how much of this speed I can sustain with more varied / tougher content...

[alantin]

@phills, that's awesome!

Really shows that it pays to vary the material and get back to your comfort zone every once in a while instead of just always pushing the limits.

 

Some work hit me in the face and it completely took all my energy to do anything extra for the last couple of weeks. Fortunately I was able to have a few chatting sessions with my tutors so the time wasn't completely wasted Chinese wise. It got easier already so now I'll take some time to recover, set goals for the next year over the holidays and hope to get back to a comfortable reading schedule again. Need to see if I can hit that 250 cpm mark by 5.5M characters too! ?

The success stories are a real inspiration when you realize that there's no reason you shouldn't be able to do it too!

[phills]

@alantin加油⛽?

 

I wouldn't have done it myself if I hadn't seen @pinion's graph and knowing it's reachable if I just put in the requisite amount of volume.  Also, without his story, I would have thought millions of chars was impossible to do in a reasonable amount of time. 

 

I got through 5.5 million in about a year, with most of it -- almost 4 million in the last 6 months.  In comparison, Pinion did 15 million in a year and a half!

 

It definitely gets easier the more you do it.  In the beginning, I resisted tracking my speed in detail, but as I got further along, I added that too, and it became its own game.  It's fun to track different factors, and see how you push your own training. 

 

Easy books to set new high water marks, and hard books to gauge how hard well known books are and to see your visible improvement over the course of a single book.  This kept me going after I got the basics of reading through a book smoothly.

 

I still have content in my pipeline I want to get to.  Plus 250cpm is a bit of a high water mark for me -- I'd like to get to the point where I can do that on a wide variety of random reading materials.  And I have quite a bit of "warmup time" that can be reduced.  (I feel an initial "wall of effort" before read anything new, which makes me want to avoid browsing random Wikipedia articles in Chinese.  I have to decide material is worth the investment before plunging in)  So more mini-games to tackle for the next year.

[alantin]

On 12/21/2021 at 6:06 AM, phills said:

I got through 5.5 million in about a year, with most of it -- almost 4 million in the last 6 months.

 

Interesting.. This can't be explained by speed increase alone. I made some projections expecting linear speed increase from my current 80cpm to 250cpm at 5.5M and spending a fixed 60 minutes per day reading would result in about 1 million characters read in the first 6 months and 1,5 million in the second six months. You must have at least doubled your time spent reading compared to the first half of 2021.

 

That seems to also speak for it getting easier to keep reading once started.

[phills]

On 12/21/2021 at 5:28 PM, alantin said:

You must have at least doubled your time spent reading compared to the first half of 2021.

 

Yea it was more interesting as I got better.  It felt less like studying.  The first half was also not continuous; I'd take long breaks between books.  While I read mostly continuously in the second half.  I had a pipeline of books already set to go.

 

Part of it also was seeing pinion's chart and deciding to go for it.  Then when my motivation started flagging, posting on this board, and turning it into a game.

 

When I was seriously reading, I'd read 2-3 hours a day.  I'd break up a book into ~25k character segments, and try to get through a segment a session.

[imron]

On 12/21/2021 at 4:06 AM, phills said:

I got through 5.5 million in about a year, with most of it -- almost 4 million in the last 6 months.  In comparison, Pinion did 15 million in a year and a half!

 

It definitely gets easier the more you do it.  In the beginning, I resisted tracking my speed in detail, but as I got further along, I added that too, and it became its own game.  It's fun to track different factors, and see how you push your own training. 

In my view, these are far better metrics to monitor than size of flashcard decks.

[phills]

After the holiday break, I'm back to my reading routine.  But not as focused as before -- now that I have reading under control, I'm switching 2/3rds of my time to listening practice.  But I'm still going to maintain a steady level of reading, and log my progress.

 

I just finished 棋王 by 阿城.  The version I read (courtesy of @publius) actually consisted of 3 novellas by 阿城 put together -- 棋王, 树王 and 孩子王.  I would rate it as intermediate difficulty for me.  The language, I think, would be considered on the "simple" side, but the setting is from the Cultural Revolution era which is not as familiar to me.  There are a lot of cultural revolution era terms that I'm aware of from previous books (like 黄金时代), but don't immediately translate in my mind. 

 

I read it at 203cpm, a solid Gear 3 speed.  Early chapters, a mix of Gear 2 and 3, and the later chapters, a mix of Gear 3 and 4.

 

chessking1: 2.52 hrs, 29687 chars, 196.33 cpm
chessking2: 2.18 hrs, 26476 chars, 202.11 cpm
chessking3: 2.42 hrs, 30760 chars, 212.14 cpm
Total: 7.12 hrs, 86,923 chars, 203.47 cpm

 

After my 3 week break, low 200s is about what I expected.  I never fully got to Gear 4 with this content.  It also took about 50k characters (halfway into 孩子王) before I warmed up enough to hit Gear 4.

 

As comparison, I read 黄金时代 (the most similar in terms of difficulty / setting) in September last year at 149cpm, so that's ~ 50cpm gain since then, across all gears.