The advice "don't drill word lists, drill sentences" is wrong

[werewitt]

For several times I've heard here a piece of advice being professed: "Don't learn word lists in Anki, learn sentences instead". With zero evidence, obviously, apart from personal 后悔. Here's a nice article from pre-Anki times, with evidence that debunks this and a few other common language-learning myths.

 

I agree that pure exposure to sentences (aka reading), maybe even drilling simple 2-3 word collocations is useful, but substituting memorising long sentences for memorising words is not.

 

Quote

Myth 1: “Knowing a relatively small number of words takes you far.”
Myth 2: “Word lists are of limited value. ”
Myth 3: “Presenting words in semantic sets facilitates learning.”
Myth 4: “Words should always be learned in context.”
Myth 5: “Words whose meanings have been inferred from context are retained better.”
Myth 6: “Words learned productively are retained better.”
Myth 7: “Vocabulary knowledge should not be tested separately.”

 

mondria.pdf

[Lumbering Ox]

I've read it here that for vocab with Chinese it is much better to go C to E rather than E to C for reasons I can't remember. It always bothered me because I've read good arguments that it is better to go productive [in a Portuguese context IIRC] and it sure seems good for Kanji.

 

Seeing that there is some research indicating that going productive is not harmful and may be helpful and sure is quicker, it eases my mind in going that route. I would however to the RTH deck productive style, writing characters at the very least would be a good thing me thinks.

 

 

As for drilling sentences, I never saw the appeal. I've seen sentences in some non Chinese decks and they make me want to stab myself in the eye with a very hot French Fry. Ditto for Heisig's idea that one should come up with elaborate stories and visualize them, I never did and just don't want to. I figure once I get vocab pounded in, I'd drill them in sentence format via native material. Good thing Chuck Norris provides us with Chairman Bao so one doesn't have to wait too long before getting your native material groove on.

 

 

But what do I know.

[NinKenDo]

I think this comes from experiences of learners who start out drilling vocabulary and start drilling sentences only later.

 

I'm actually a very strong advocate of sentences over vocabulary, and I would say I still am. But having started Chinese with sentences from the outset, I'm actually having a lot of problems, and have begun working on a way to integrate some effective vocabulary flashcards into my study repertoire again.

 

I think it's easy to hut a wall with vocab cards and start to see heavily diminishing returns later in the language. To then discover sentence methods, and then suddenly see a lot of gains from using them. And conclude that sentences are the superior method. Rather than recognising that sentences were only so effective because of your prior vocabulary drilling.

[werewitt]

51 minutes ago, NinKenDo said:

I think it's easy to hut a wall with vocab cards and start to see heavily diminishing returns later in the language.

This is the common (pre-)intermediate motivational problem - when one is a beginner and knows 100 words, a daily 10 words is a massive 10% increase (hurray! ).

When you know, say, 3000 words, the same daily amount of words - 10 - and presumably the same amount of work - results in a miserable ~0.3% improvement () which doesn't expand the range of accessible materials by much.  

 

Nothing to do with vocab cards themselves, purely a motivation trouble. I suspect that a later "discovery" of sentence method by some just this - a new stimulus for drilling, and not a superior method of.

[imron]

The problem is not so much wordlists, it's wordlists that contain words without context.

 

Sentences solve that problem to some degree because they have context.

 

If you were studying word lists that you have generated yourself from content that you are reading, then learning from wordlists probably will be quite effective.

 

If you're just going through HSK 1-6, or some other frequency based list devoid of any context (e.g. it's effectively random order) then there are better ways to go about it.

 

I know you might think - that's Myth 4, but actually it's not.  I'm not saying to *always* learn words from context, just that learning from context will be more effective most of the time (context gives your brain a memory hook).

[werewitt]

8 minutes ago, imron said:

just that learning from context will be more effective most of the time (context gives your brain a memory hook).

Arguably, more information to remember for one item limits the number of items that can be processed by said remembering mechanism

Perhaps having sentences as illustrations on a vocab card, that would probably help. But not memorising production or recognition of those sentences.

[艾墨本]

First, some sarcasm, because why you gotta open with an attack on another person's opinion?

*clears throat*  I'm sorry you've encountered such a horribly incorrect statement on these forums such as a recommendation to study sentences over words. I feel saved.

 

Reading through the article, the advice it gives mirrors the advice given on these forums repeatedly. For example, see what imron has said above and what the article says:

Quote

The fact is that word lists in principle contain the most frequent words from frequency lists, lists in which the words have been ordered on the basis of their frequency in actual language use

But, as imron often points out using his CTA tool, the word list that the HSK provides is often inadequate, which the article supports here:

Quote

At this [80%) level of text coverage, there is no adequate text comprehension yet. Only at a text coverage level of 95%, the majority of the readers will understand a text reasonably well

 

And goes on to say:

Quote

For such a text coverage, you need a knowledge of 3,000 to 5,000 word families, a word family being defined as a base word with its inflected forms and transparent derivations

But, assuming you know all of HSK 6, which takes us to the upper limit of this approximation, you are often left with only 80% comprehension, or less (at least in my tests using CTA), which begs the question of how effectively the HSK word list was created and what corpus it stems from.

 

Now another quote from the article:

Quote

Incidentally, this process of ‘contextualizing-decontextualizing’ can be excellently realized with the help of word cards, namely by putting on the front the word to be learned (e.g. une canne), and on the back on top the context sentence (e.g. Le vieil homme marche à l’aide d’une canne.) and at the bottom the translation (a walking stick). In this way, the learner has the context at hand as a cognitive foothold while learning, and at the same time, he works up to knowing the meaning of the word without any context. Thus, context helps in learning, and at the same time, it becomes clear to the learner that words should not always be learned in context.

And, coincidentally, a picture of how my Anki cards are set up:

 

1 hour ago, werewitt said:

Perhaps having sentences as illustrations on a vocab card, that would probably help. But not memorising production or recognition of those sentences.

Agreed. Though sometimes memorization comes as a byproduct, and other times I just love the sentence, like: 做牛耕田，做狗看家，做和尚化缘，做鸡报晓，做女人织布，哪只牛不耕田？

 

This article further mirror's this forums love of SRS: 

Quote

I suggest the so-called ‘hand computer’ (Leitner, 1972; Mondria & Mondria-de Vries, 1994): a deck of cards with a sophisticated repetition system on the basis of ever bigger intervals, guaranteeing optimal long-term retention (see Figure 2). And those who prefer a computer program based on the same principles, can find all kinds of information at: http://de.wikipedia. org/wiki/Lernkartei-Software. 

 

I found this segment particularly interesting:

Quote

The learning effect of inferring per se is rather limited: after two weeks, only 6% of the inferred word meanings were remembered. The addition of a verifying stage led to an extra retention of 9%. However, it is only when the word meanings are intentionally memorized that the learning effect becomes substantial, as shown by the retention figures of the meaning-inferred method (47%) and the meaning-given method (50%). What is most striking is that the meaninginferred method (inferring + verifying + memorizing) does not lead to better retention than the meaning-given method (memorizing the given meanings): the level of retention is similar. Thus, in this experiment no evidence can be found for the idea that inferred word meanings are retained better.

 

5

Essentially, learning cloze sentences in Anki is not any better than just learning the words as pictured on my Anki card above, so long as it is paired with the "intentional" aspect of including the meaning. I think this is what is usually done, with the cloze on one side, the answer on the other, and a definition below

 

[艾墨本]

My post is too long and started crashing my browser, so I can't edit it.

 

Continuing here:

 

But at the end of discussing learning through inference he says:

Quote

Does this imply that learners should not infer word meanings from context any more? Of course not, as inferring is a useful compensation strategy when our vocabulary knowledge is limited. However, inferring is not the most efficient learning strategy.

1

Cool. 

 

All said and done, I can't take this article with too much weight as he states several times that his research involved experiments with his own students. This hardly sounds controlled and is certainly a small sample size that can't account for cultural differences in language aquisition. You can find many articles on this topic here: http://www.ameprc.mq.edu.au/docs/prospect_journal/volume_21_no_3 I jump around the journals by swapping out the numbers at the end of the url. The link is an arbitrary bookmark.

 

So, is he right and is everyone who suggested using sentence memorization as a learning tool wrong? I don't know, and nor do you. What you do know is if one method works for you or doesn't work for you, which is far more useful. He even concedes that individual preferences can have an impact on what methods work when he states: 

Quote

First, they are not particularly interesting to learn (although this argument does not hold for certain types of learners)

 

[werewitt]

26 minutes ago, 艾墨本 said:

This hardly sounds controlled and is certainly a small sample size that can't account for cultural differences in language aquisition.

All of the advice given here to the contrary lacks even that minimal level of support. It is all based on personal opinions and nothing else. Am I mistaken and there was something here supported by a double blind controlled experiment on a large pool of learners that proved sentence drilling is superior to plain vocab learning?

[Guest realmayo]

Personally I think the word-list-phobia on these forums actually stems from how people deal with characters.

 

There's a lot to be said for binge-memorising a load of characters at some point. If you can already speak some Chinese but can't read any characters, then a couple of months memorising characters and boom! you can suddenly read some Chinese.

 

But I don't think binge-memorising a vocabulary list has the same result. Probably because memorising a character as symbol is much simpler than memorising a word as word. And anyway you need to learn 5x or 10x more words than characters.

 

So people on the forums advise others not to use the characters-approach for learning words, which is sensible.

 

However it depends on your level too. If you know, say, 10,000 words, then your vocabulary will only increase really slowly if you insist on only learning words which you know you'll  encounter repeatedly in context. Instead you're more likely to come across a word, note it down as worth learning, and later learn it, along with a bunch of other words picked up in similar circumstances. That's not much different from learning a word list.

 

Also, for learning words in context:

Arguably, these words are in context:

 

老虎 tiger
老鼠 mouse, rat
鼠标 computer-mouse
松鼠 squirrel
松树 pine tree

[Guest realmayo]

Also I'm not sure I've seen anything like a consensus on these forums that the best way to learn vocabulary is by drilling sentences. I think that's firmly a minority view, no?

 

My guess is it slows down the learning of vocabulary. It may mean you know the words better. But if you learn words faster you get to read more and the more you read the more you see words in context anyway. Not clear how that trade-off works out.

 

However if people are memorising sentences, that's not just going to help them with vocabulary, but with lots and lots of other good things too (assuming the sentences are good, natural ones).

[Guest realmayo]

And I think memorising texts is a great thing. It's one of those traditional learning skills which every civilisation has encouraged students to do. Western society has decided over the last 50 years that it's a bad thing. I'm not sure why. But it seems to me that memorising sentences is just a welcome throwback to the traditional, time-honoured method of learning by memorising texts.

[Wurstmann]

I think it also depends on how you use Anki. I use it to get an idea of a word and to remember the reading. I read the sentence and if I can understand what it means I press 'good'. Later, as I hear and read that word in more contexts it gets more and more defined in my mind.

My cards look like this (back, front only has the sentence in Chinese and without bolding):

[eddyf]

1 hour ago, werewitt said:

All of the advice given here to the contrary lacks even that minimal level of support. It is all based on personal opinions and nothing else. Am I mistaken and there was something here supported by a double blind controlled experiment on a large pool of learners that proved sentence drilling is superior to plain vocab learning?

Get used to it... it's a web forum, not an academic journal. Besides, there's far from an academic consensus on the best way to learn a language. How would you even design a blind experiment to compare language learning strategies? It doesn't make any sense.

 

Maybe it makes you feel better to think that your way of studying has a strong scientific basis. Ok, sure, knock yourself out. The reality is that no matter how you study, it will still take thousands of hours to learn everything. So it's more important to focus on keeping up your own motivation, and maybe making the journey a bit more enjoyable along the way.

 

As to the topic at hand, I think the majority of people do use word list drilling, often exclusively, and the advice to drill sentences instead is a reaction to that, because exlclusively drilling single words leaves huge gaps in your language knowledge which drilling sentences can partly make up for. There's a place for both of them and which of the two is more worth your time is highly dependent on your level and what language competencies you're trying to focus on.

[werewitt]

1 hour ago, realmayo said:

Also I'm not sure I've seen anything like a consensus on these forums that the best way to learn vocabulary is by drilling sentences. I think that's firmly a minority view, no?

Well, here everyone was "sentences or nothing"  If you mean it's a sensible course of action to ignore any advice given on this forum, I agree. After all, everyone eager to give this advice seems to be an amateur and the % of people here who actually learned the language to an OK degree is infinitesimally small.

[Shelley]

Interesting discussion. I don't learn sentences, I do learn word lists but not in isolation, the lists I learn are the 生词 list that come with each lesson in NPCR. So they have context in the lesson and example sentences in the substitutions in the drills and patterns section.

I didn't realise people learnt whole sentences by rote, seems a bit clumsy to me, learning patterns and how to use them seems much more useful. i thought the sentences with new words were for examples of how they were used and not intended for memorisation.

 

I don't think learning long lists of words/characters just for the sake of it is useful, unless for example you were interested in a particular subject and wanted read about it or other similar usefulness, but I would think this was for advance learners who's grasp of the language was very good and just wanted to add vocabulary.

 

Everything in moderation, don't overthink everything, learn the best way for you and share things.

 

1 hour ago, werewitt said:

If you mean it's a sensible course of action to ignore any advice given on this forum, I agree

 

You have just agreed with advice given on this forum.

 

This is a place full of people learning chinese, why do think anyone will be anything but an amateur? What do you expect of people at all levels learning a language. If you really crave "professional" advice/teaching pay for a highly qualified teacher/tutor. There are however some very highly advanced learners and teachers on this forum. If you hang around long enough and are pleasant you may get to meet some of them

 

 

 

 

[querido]

I never said "memorize sentences".

And, it is generally understood that memorizing sentences is not the point even in e.g. Glossika threads.

**

Flashcards having an L2 sentence or more on the front with a cloze deletion, and no English on the back either, let me spend my flashcarding time in L2; I don't think that staying in L2 is bad for me. If that is bad for me then the problem is flashcarding, as I've suspected.

[querido]

2 hours ago, werewitt said:

Well, here everyone was "sentences or nothing"

You misrepresented what they said.

[陳德聰]

Yeah I can't believe I just wasted my time reading that thread only to find that it was another case of you not reading the thread...

 

I didn't even realise people did cloze or sentence cards exclusively. This whole thread seems like a big straw man for personal satisfaction?

[Guest realmayo]

1 hour ago, 陳德聰 said:

This whole thread seems like a big straw man for personal satisfaction?

 

Anonymous person feels the need to belittle strangers online... probably not the first occurrence in the history of the internet.

 

7 hours ago, Wurstmann said:

 

I think the reason I'm scared of using that kind of sentences-method is because I'd fret that my sentences weren't natural Chinese. For some reason, that sentence about the 绝命书 makes me uneasy, almost certainly it's the fault of my limited sense of when 发现 is most likely to be used. But for me, personally, it would create too much background noise in my head each time I saw the card: is the sentence okay? under what circumstances? should I double-check? does this mean I need to do some more work on 发现? etc etc.

 

Actually, I'll hit LINEdict or other resources for example sentences when I learn new words and put them in SRS, but I only put these sentences as 'additional information' in the the answer field. Even then I'm not entirely comfortable. I'm tempted to instead do one of those massive mining exercises people have done on subtitled movies or TV, guessing that those sentences are likely to be less problematic.