Memorizing vocabulary at the advanced level

[songlei]

it goes without saying that the acquisition of vocabulary is one of the cornerstones of learning a new language, and chinese is no exception. however, i have been doubting the merits of memorizing vocab at the advanced level, at least the level on which i am now. back in the day, i used to learn words such as 已经，走路，汽车, etc., and moving further up the ladder learning things such as 政权，反道德行为，冷漠，and so on. what all these words up to a certain point have in common is that they are used with some frequency, and are thus fairly easy to remember. but i now find myself in a very problematic state. when i read a novel or a news article that talks about some specific event (e.g. a fire, not politics or economics), i am still confronted with massive amounts of new vocabulary, but only of the sort that you rarely see. the type of vocabulary you will find used only once in an entire novel, and probably not at all in the next you will read. but the vocabulary of this type does still take up about say 2-5% of the texts you will encounter. it appears that the road to tackling this 2-5% is the longest and most difficult, since the reservoir of rare and advanced vocabulary is the biggest, and the chances of each individual word or expression returning the smallest. because of the low frequency of encounters with these words, combined with their complexity (they are mostly words and expressions that we don't use in our own languages) as well as the vast amount of new ones you will encounter during a reading session, that is if you read at a relatively fast pace, it makes it almost impossible to master them, even with tools such as srs. i was wondering, what do the experienced students of chinese among you do in order to overcome this problem? is memorizing vocabulary still a major element of your study routine? or do you focus on individual characters more so that you can infer the meaning of new combinations based on your knowledge of the more limited amount of individual characters that exist? i'd love to hear what those who are more experienced than i have to say, as i am quite at a loss as to how i could proceed more fruitfully with my study. thanks and hope to hear from you soon!

[creamyhorror]

I was thinking about this problem not too long ago. I have the impression from reading an article here and there that Chinese writing actually has a greater proportion of "in the middle" words that aren't common but also aren't rare enough to ignore. (See the Zipf-Mandelbrot Law.) Basically it's easy enough to learn enough words to obtain coverage of 80+% of texts, but beyond that the number of words you have to add on to gain another percentage point of coverage starts going up faster (and all those unknown words really get in the way of your reading comprehension). I think for English the curve might not be as steep.

Anyway, I think at a certain point it's more effective to just read like mad and let unfamiliar words just soak in through sheer repeated exposure. If you have problems with any in particular, you can put them into an SRS for reinforcement. Don't bother memorising or SRSing everything you come across; just let the most commonly-occurring ones soak into your memory via osmosis. I think it's more time-efficient to do that.

I also think knowledge of characters comes into play - if you know 1 character in a 2-character word, you can guess the meaning of the word (since the characters are often synonyms). I often check the meanings of unfamiliar characters for this purpose - when I see them again in an unfamiliar word, I can make a good guess at the meaning.

Hope the advanced learners and native speakers can chime in on this issue!

[kdavid]

Before I chime in, I need to explain something I've noticed when it comes to language learning: the difference between recalling and recognizing.

When we recognize a word, the word we recognize could be part of either our passive or active vocabulary. High-frequency words--those that we see everyday--are likely to be part of our active vocabulary--meaning we use them on a day-to-day basis. Such high-frequency words include common adjectives, household items and appliances, etc. Low-frequency words are words we don't use often, if at all. Such words could include sledge hammer, asphyxiation, NyQuil, etc.

When we can recall a word, it means that as we are speaking or writing, we can actively produce the word in context without "searching" for it. If you can recall a word, you can also recognize it. On the other hand, a word we can't actively recall, but we know "on the tip of our tongue" is more likely to be a word we can only recognize.

I want you to think about a specific movie star affectionately called (at least for now) "The Governator". He's starred in tons of action movies like "True Lies" and "Predator". Now, I want you to spell his full name from memory. Can you do it? Probably not, and the fact that you can't spell Schwarzenegger's name from memory, but that you can read it and identify it to a specific individual means that his name is part of your passive vocabulary, not your active. In other words, in this case of spelling, "Schwarzenegger" is something you can recognize, but not recall.

Now, when it comes to learning vocabulary, there are tons of words which are of questionable usefulness when it comes to pounding them into your active vocabulary. Do you really need to know a dozen different synonyms for the word "grief", or how to say all of Hollywood's stars names on the spot? Probably not. However, should you be able to recognize these words when you come across them? Sure, why not.

As such, it's important that as you're learning that you're deciding which words/phrases you want to be able to recognize, and which ones you want to be able to recall. For words which you only need to recognize, you can simply make one card which tests you on reading the characters for that word. For words which you want to be able to recall, you make two cards: one testing your ability to read the characters, the other testing your ability to produce the word with a simple prompt. Such a card for "Schwarzenegger" may prompt with: Former governor of California - famous action movie star

In short, a lot of the low-frequency words you see out there don't really need to be part of your active vocabulary, you only need to be able to recognize them when you come across them while reading. However, if you really want to sophisticate your manner of speaking, learning how to properly use low-frequency words can be useful in making you sound like a pompous ass, which I like to do quite often just to show off.

[creamyhorror]

Illuminating reply there, kdavid. Sometimes I try to move words and expressions from my passive to my active store, but it rarely works. The ability to correctly and elegantly deploy rarer words and fancier expressions in writing generally impresses me ...while in speech it might strike me as bombastic.

Anyway, for normal functioning in society I certainly agree we can stick to learning to recognise words rather than recall them. Sometimes I browse the dictionary and end up adding words to my SRS that I don't even know if I'll ever see in writing, and regret spending brainspace/time on them.

[Guest realmayo]

Very good question songlei, one I have been wondering about. I reckon there's about 1000 words in the HSK vocab list I have yet to learn, though there are plenty outside the list that I know. I use SRS every day, and until recently I was looking up words I didn't recognise and adding them to the SRS. But now that I've started really reading lots (mainly magazines) I'm getting way too many new words to add to the SRS: sometimes these are words whose meaning I can guess the first time I see them (given the context, and knowing the two characters that make up the word).

What I've ended up doing is work out that at the moment, learning 30 new words a day is a reasonable quota for me. So, the first 30 I come across (excluding some which I think are too rare to bother learn at this stage) I add to the SRS, the rest I don't learn and, once I've got my quota for the day, I might not even bother look up words when reading certain articles (ie if I'm practicing improving my reading speed).

The alternative, of earnestly underlining and then looking up and noting down and adding to the SRS ... was just overloading me, and I was spending much longer on vocab than on reading, which can't be right.

And I have found that, while SRS is good in its way, if there's a not-so-common word that I learned months and months ago and added to the SRS, and then see for the first time in a magazine, and remember it correctly, then it's at this stage that the word really seems burned into my brain, and I'm confident I'll remember it for a good long while. Which just makes it seem more obvious to me that I should never let the time I spend learning vocab get in the way of time spent reading.

But I'd definitely be very interested in hearing from others at a similar stage, and those who are some way beyond. (I should add, I'm talking here more about 'passive' than 'active' vocab.)

[c_redman]

I'm also at a similar stage. I can read novels, although slowly, and know about 90% of the words in a chapter, when I count them. At this point, the unknown words are of such low frequency that I'm not enthusiastic about learning them. But what I have done just recently is to focus on characters more closely. I am using Skritter to finally learn to write the top 3000 characters. This focuses more attention to the components of the characters, and actually increases the speed of recognizing them for reading. Knowing words is preferable to knowing the component characters, but after knowing 8,000 words or so, studying the next 5,000 low frequency words may not help reading ability as much as learning 1,000 unknown characters which can at least give you a chance of guessing the meaning. Someday, I'll do the number-crunching to see if this is actually true.

This is really just a short-term fix. In the long run, there's no avoiding the need to learn more words to improve fluency. This is the same challenge in learning any language.

creamyhorror, the Zipf plots are similar for both English and Chinese. There is a slight difference at the lower frequency end, but with this log-log scale, that means word 10,000 is 5 occurrences per million versus 10.

[creamyhorror]

It adds up, though. With 5,000 English/Chinese words in your vocab, what's the percentage coverage of the LCMC versus the BNC? A few percentage points would constitute a big difference.

[songlei]

how nice to see so many interesting reactions already. it seems as if this is a commonly perceived problem. c_redman, i don't understand much of that graph you attached, but do i understand correctly that you're trying to measure the distribution of the frequency of certain words in languages? according to which standards are they ranked? i too have a hunch that at a certain point, an increased focus on individual characters becomes more feasible in the light of the war against numbers. in other words, if you have to choose between learning an extra say 30,000 words vs 5,000 more individual characters so that these new words become understandable in their context without the need for a dictionary, which option is more efficient? to me, it seems that the second approach is a more intelligent process, and therefore more promising in terms of building up a flexible understanding of how chinese is written and how meanings are composed, whereas the first is faster but more tedious. of course, there is no way of measuring this, but perhaps there are those who have experience with either two methods?

[c_redman]

Ah, I see it now. If you knew the top 5,000 English words (lemmas are a better comparison to Chinese, but the list I have only goes to 6,300), you would know 89.5% of the BNC, but you would need to know 9,010 of the top Chinese words to know the same amount!

To know an extra 1% of the respective corpus (90.5%) by knowing the next most common words, you would need to know 809 additional English words, but 957 additional Chinese words. The amounts gained are similar because the frequency for English words around rank 5000 are about the same as Chinese words around 9000.

This is all simplification, since nobody knows just the top N words and nothing else; it's a probability distribution, so that there's a chance of knowing even rare words. Also, the ability to "know" words by context or by guessing from the component characters is a factor. In addition, one can gain a specialized vocabulary in particular domains, so that you can be quite good at reading newspapers but struggle with novels.

edit: songlei, the word count of every word in the corpus is the basis for the rank. So, 的 is rank 1 with frequency 61,778 per 1 million words, 了 is #2 with 15,447 per million, etc.

Edited November 17, 2010 at 11:50 PM by c_redman

[imron]

In my own experience memorising words and characters at this level is essential if you ever want to see improvements in your Chinese. Otherwise you'll just coast by on existing skills, and get stuck in the rut of always only being able to read 95%.

At this stage, learning by osmosis is useful, but also full of pitfalls because many words that you think you can guess the pronunciation/meaning of, you actually find out later that you were wrong (秀才识字识半边 and all).

Also, don't worry about words you think are too uncommon that they'll never appear in other novels. At the beginning of the year, I decided I wanted to really focus on my reading, and since then I've gone through about 14 novels (it's amazing how much spare time you have when not reading the forums ). I'm constantly surprised at how words I thought I'd never see again continue to pop up over and over, both in the same novel and in others. It always makes me happy and gives me a sense of accomplishment when I see a word I learnt in one novel appearing somewhere else.

To begin with, personally I think the most important thing to do is develop a reading habit. You need to make it so that reading is not a chore. Despite previously already being able to read quite well, I found that reading for extended periods of time (1/2 an hour or more) was always painful and it was easy for me to get distracted. Since starting to make an active effort to change this, it took about 4 novels before reading for long periods (1/2 an hour or longer) became pleasurable and not just something I was forcing myself to do. It probably took a further 4 novels after that to get to a point where I could be engrossed enough in a book that I wouldn't notice that 3-4 hours had passed.

In order to accomplish this, you need set yourself a target of reading at least 1/2 an hour a day and stick to it no matter what. Whatever happens you need to make the time for this, and you need to keep at it every day in order for it to become a habit. At the beginning it will be painful and slow, but it will get easier the more you do it, especially as you will be learning new words as you go and this in turn will make future reading easier and so on. Please bear in mind that any memorising of words/characters needs to be outside this 1/2 hour. That is, don't spend 10 minutes reading and 20 minutes looking up words, you need to make sure you're putting in the reading time. Realistically speaking, this means you may need to set aside at least an hour a day - 1/2 an hour for reading, and another 1/2 hour for learning new words and revising.

Something like Pleco is an essential tool for this sort of activity, because it allows you to look up words quickly as you are reading and add them to a list for later revision. This means you have minimal interruption when reading, the ability to understand what you've just read, and the ability to then revise newly encountered words at a later point in time. If you don't have Pleco, underlining new words with a pencil is also a good way to achieve a similar thing, although going back and looking up those new words is more laborious.

When doing this there are two things you should pay attention to. Firstly, you need to be reading things of the appropriate level. If you start reading something and there are too many new words it will mean constant interruption to the reading process, which will make things less enjoyable, and also give you the feeling that you are making no improvements. If on the other hand you pick something that is more suited to your level the opposite will be true. Case in point, at the beginning of the year, the first book I started reading was 金庸's 书剑恩仇录. Although I could read it, there were too many unknown words per page to strike a comfortable balance between reading and learning new words (see below about quotas). It's not that I couldn't have read it, but more that I wanted to focus my time more on reading than on learning new words, and so 10-15 pages in, I put it aside in favour of something easier. Fast forward 9 months, 10 books and some 2 million characters later, I picked it up again and this time was able to breeze through it. This was partly because I was now much more comfortable reading for extended periods of time and partly because I'd already encountered many of the words that were previously new to me in those other novels, but spread out over time and so the frequency of new words was greatly reduced. It was really good to do something like this because it provided great positive feedback showing me the progress I'd made over the year.

This then leads in to the second thing to pay attention to, and that is to set yourself a sustainable quota of new words to learn on a given day. For me, I've found that this is 5-10 words a day (with a word sometimes containing more than 1 new character). I could learn more, but then I'd find I'd be spending more of my time learning and revising new words than on reading, which is not something I want to do (I prefer to get the revising through more reading).

When doing this, you'll then need to prioritise which new words will fill up your quota, and which words you can leave to another day. Don't worry that you might be missing out on learning important/useful ones by not learning all the new ones at once. The important ones will keep showing up in other places so they'll fit in a later day's quota. The less-useful ones wont repeat and are therefore probably not worth your time learning.

I tend to prioritise in the following order:

1) Names of people/places - These take top priority for me because in a given novel the names will be repeated often enough that it should be quite trivial to keep them in memory. Common characters used in names/surnames will also continue to pop up in different variations in both the same novel and others, plus in newspaper articles or anywhere else you might expect to find a name, so they're always useful in that respect. I typically don't place too much emphasis on learning the meaning of the rarer characters used in names, more just the pronunciation. Also of note, is that it's worth looking up the pronunciation of surnames even for characters you think you know, because quite often the pronunciation of a character when used as a surname is different from the pronunciation you might be more familiar with e.g. 单 Shàn, 曾 Zēng, etc.

2) Words that look suspiciously like words I already know how to pronounce. At higher levels, I've found more often than not that less-common words containing parts of characters I already know typically contain different pronunciations to what I'd expect e.g. a couple off the top of my head 倩 qiàn, 栉 zhì, 耿 gěng. If you don't make sure to learn the correct pronunciation, you'll fall in to the trap of guessing the incorrect pronunciation and then start to reinforce that mistake each time you see the character, which will make it that much more difficult to correct later.

3) Words that I've seen previously, but that I didn't learn because I'd already filled up my quota for the day. Logic dictates that these words are more common (at least in the context of whatever I'm reading at the time) and therefore more likely to be worth learning ahead of others.

4) Other new words.

When learning words, in general I'll read the meaning to get an understanding, and then relate it back to the part of the novel where I read it. I typically won't spend much time trying to memorise the meaning. If the word is useful it will continue to come up in other places and that will help reinforce the meaning through context. In terms of memorisation, I'll really only focus on the pronunciation.

Anyway, this is now starting to get a little long so I’ll finish off by saying the most important part to all of the above is perseverance, and making sure you keep reading a little bit every day. You can memorise all the characters you like, but it’s not much good if you find you can’t then read more than one or two pages of Chinese without getting tired/distracted and switching to some other activity.  You can use tools like "Don't Break the Chain", or my own 100% to provide motivation and keep track of progress.

[bande]

I really enjoy the direction that this discussion is going in. I've been pondering similar thoughts for the last year. I guess the simple but difficult answer is too read as much as one can, and the that's the strategy I've had the most success in following. I find that it's not too difficult to pick up an approximate meaning for a word through context, but it's the lack of a pronunciation for a character that's the most jarring. So in terms of dictionary and SRS use I find it more fruitful to focus more on pinyin readings than on definitions since definitions for the most part can be easier learned through reading. A further strategy based on this which I've contemplated is something like the Heisig book two and just learn the readings for say the 3,000 to the 5,000 most infrequent characters.

[Guest realmayo]

The talk of percentages and corpuses and so on, I just have a feeling that while this might possibly have some use towards a quicker and more efficient learning of beginner-vocabulary, it gets a lot less useful as the percentages get smaller. You've also got to bear in mind that it can be much easier learning a brand new, rare, character which only appears in one word (say, 隼 = falcon) than remembering something like 同为 where both characters are very common and appear in lots of other words you already know but the meaning is not necessarily immediately intuitive, and therefore easily (at least in my case) forgotten.

Imron, I'm happy that my current way of doing things is similar to what has worked well for someone else, it means I can relax a bit about not learning all the new words I come across. Though if you've written about the quota idea before in the forums then I've almost certainly read it so perhaps that's why.... I've got a couple of months of just studying Chinese which is why I've ramped up to 30 a day.

One other small thing involves 成语: I hadn't really bothered with them until recently, but now I've started to learn some of the more common ones, I'm surprised about how often they crop up. Sure, sometimes it's obvious that a 4-character combo you don't understand is a 成语, but often they just seem to me like two sets of adjacent 2-character words which I don't really get ... until I realise it's a single piece of vocab I need to learn and recognise in the future.

[valikor]

What I've most recently been doing is reading a book and recording every single new word and Chengyu, even if I was able to easily guess. Then I ask a native speaker if there are any words on the list that are simply too rare or obscure that I shouldn't even bother right now (This gets rid of about 15% of the unknown words)

I then put everything into anki and have it add 40 words a day.

I also ask my native speaker if any of the words are especially important, then I can flag them in Anki so I know to find some example sentences and learn how to (hopefully!) actively use them.

This seems to work pretty well. We'll see in a few months.

[imron]

I envy people that can learn that many new words in a day. I've always failed miserably when setting myself high rates, and often started out strong but then before long burnt out and wouldn't study much of anything for like two months :blink:

I'd certainly be interested in hearing back in a few months about the results, both in terms of being able to maintain that rate, and how effectively you feel you have learnt all those words.

I've posted previously about my belief that more is not always better. I wonder how that other poster is going now.

[anonymoose]

A previous poster mentioned active and passive vocabulary. I think actually vocabulary can be divided into several layers. There's the everyday vocabulary that you can use without even thinking about it. Then there's a level below this of words which you can recall fairly quickly and use naturally, and below this, a level of words that you know, but have to struggle for a few seconds before you can recall them. I guess these all these belong to active vocabulary. Passive vocabulary to a certain extent overlaps with active vocabulary, in the sense that the bottom layer of active vocabulary merges with the top layer of passive vocabulary. Then in the lower layers of passive vocabulary, you have those words that you can't recall but understand when exposed to, and at the bottom, the words that don't mean much when you see or hear them, but you have an inkling that you've learnt them some time a few years ago.

My problem is that my passive vocabulary is quite large, and bottom layer of my active vocabulary is also fairly broad. But I'm finding it quite difficult to move vocabulary into the top layers of spontaneous recollection. Therefore, whilst my reading is quite good, and writing also not bad, my spoken Chinese is nowhere near the level I would like it to be. After having lived in China for over 4 years, I would have hoped to be able to speak fairly fluently on a deep level. On some days I find my speaking is not bad, but on other days, I still get stumped by even the most basic phrases when I need to say them out spontaneously. I'm not really sure how to overcome this, except just to keep on practicing.

As for reading novels, that is undoubtedly a good way of broadening vocabulary, but the precondition is that you enjoy reading. Personally, I'm not a big fan of reading novels, even in English, and therefore reading novels in Chinese is a bit of a chore for me, and not just because of the language.

[valikor]

Well, I am essentially a full-time student (I just work 15 hours a week), and I live in China, so I don't think 40 words a day is really that many.

Unfortunately, to be honest, this learning is largely passive. In the past few months I have not spent as much time as I should having Chinese conversations, though I've spent lots of time reading and SRSing.

I've increased by reading level somewhat quickly but still have a lot of problems activating my vocabulary when writing or speaking. I assume this has more to do with neglecting the exercise of active skills, and less to do with how many words I am studying daily.

I would expect this problem is not uncommon. I'm sure the simple/not-so-simple answer is to just practice more.

EDIT: Looks like anonymoose beat me to it, and what he wrote is probably more useful

Edited November 18, 2010 at 02:14 PM by valikor

[creamyhorror]

The talk of percentages and corpuses and so on, I just have a feeling that while this might possibly have some use towards a quicker and more efficient learning of beginner-vocabulary, it gets a lot less useful as the percentages get smaller.

Really, the point behind my bringing up the frequency distribution for Chinese words was just a linguistic question. I was wondering why Chinese seems to require speakers to know significantly more words than English speakers to achieve the same level of corpus coverage. On the face of it, this indicates that Chinese vocabulary is just plain onerous to learn :unsure: But it could also be the result of numerous cross-links between words with overlapping characters that make word learning easier (to a greater extent than English word families simplify English learning).

c_redman's statistic that 89.5% coverage is achieved with 5,000 words in English versus 9,010 words in Chinese is surprising, if accurate. The gap's even bigger than I expected. And I expect it widens as you move further up the % coverage scale. If the stat is based on a fair comparison (lemmas to lemmas), then if you get to an advanced level in Chinese you may end up knowing even more words in Chinese than in your native language (and still not be more literate in Chinese versus the latter).

[ChouDoufu]

This is a great discussion.

I wanted to address one point that @realmayo made:

The talk of percentages and corpuses and so on, I just have a feeling that while this might possibly have some use towards a quicker and more efficient learning of beginner-vocabulary, it gets a lot less useful as the percentages get smaller.

While I agree that frequency data or percentages aren't extremely useful. Corpora can be useful in lots of different ways. For instance, a corpus could tell you whether the word you just read in that book by 老舍 is still commonly used or not. Additionally, a good corpus can provide a large number of examples and collocations for a word, which is extremely helpful when trying to better understand when and how a word is used. I fully expect corpora will be a key component of the next set of advances in language learning.

Unfortunately, there aren't any easy to use corpora for learning languages (let alone Chinese). If you're interested in seeing the different things a corpus can do, I recommend getting trial access on SketchEngine, which has one of the best selection of Chinese corpora. The website Wordnik is also doing a lot of corpus related work for English.

[Guest realmayo]

Volume of new words: I put them into my SRS once I've learned them, and while i'm doing that I copy that data into a spreadsheet ... which informs me that over the last 6 months I I've been averaging 20 a day, though that includes individual characters. In the last 3 months it has risen to 35 a day: at one point I tried doing around 50 but it was way too much and yes, I think took a week off following that meltdown! And Imron, I'm sure your vocab is much more extensive than mine, so you will see recently-learned words less often while reading than I will, given that I'll be including some more common words than you will.

Effectiveness: as I say, I use SRS to help me remember, but I often fail a bundle (maybe as much as a quarter? though that sounds too much...) at about the 1-month mark; however, with SRS if you fail a card once you start with it from new, and I think that the huge majority of these 1-month-failed cards I am then able to relearn well enough so they stick in my brain a great deal more firmly the second time around.

It looks like I'm spending around 90 mins every day on new words: that includes learning, testing (SRS) and admin (ie putting them into the SRS, finding example sentences for them and so on). I won't have time for that next year, which is why I'm trying to make the most of the next couple of months.

90 mins every day does seem a big committment, but it equals 1000 words a month and it is only for a fairly limited time. I think I'll be halving it early in 2011.

I imagine there are plenty of people who would laugh at all this heavy review of vocab, and advise more and more reading instead but I'm fairly happy with recent progress and, as I say, I am reading lots too. Though I do have my own doubts occasionally....

[taijidan]

Thanks to songlei for starting the thread and that was a great post from Imron on building up the reading habit.