Which is more effective for vocabulary learning with flashcards: translating from known to target language or from target to known language?
This mightn't be more 'efficient' but it's probably more effective - Go monolingual, using the TL on the front and the TL definition on the back.
They are doing two different things:
From target language to native language is teaching you passive vocabulary recognition.
From native language to target language is teaching you active vocabulary recall (useful for speaking).
The SRS I use quizzes in both directions when the sentence or word is first introduced. All subsequent quizzes involve producing the L2 word or selecting the appropriate cloze L2 word to complete a sentence.
Just my personal preference, but I'm also a big believer in sorting my available reviews by Descending SRS order.
That way, when I'm starting a review session for the day, I'm beginning with the stuff I know the best. If I feel I've mastered it, I hide the card, and it's never seen again. If not, a do the review and I'll see it again in 60 or 180 days or whatever cycle it's in.
Especially in a more challenging language, there's something motivating, non-demoralizing about starting with the most well-known sentences first. I think being exposed to a lot of medium-known to well-known stuff that's about to be mastered gets the Japanese motor in my mind started and provides kind of a sticky knowledge base for other vocabulary to stick to as I finish working through the rest of the more difficult review cards that I have for that day.
@kimojima: I'll try that tomorrow! Usuallly I do it the other way around, because it is no problem to postpone the words with big intervals for a few days, and until now I did not have any problems with it.
But I never thought about the motivation factor.
I don't know if I'm the right person to get involved in yet another SRS thread, because I don't use the technology myself. But from what I hear, the best way is not to use translations at all, but instead to keep everything in the target language. This obviously doesn't work for beginners but might be interesting at the intermediate stages.
In my experience, monolingual dictionaries tend to be more accurate and nuanced than bilingual ones. Obviously you want a really verbose one that explains the word or expression and may even have example sentences, not just a list of synonyms. These explanations are often especially helpful for understanding idiomatic expressions and colloquial language, which can be difficult to translate into a second language. Keeping everything in the target language naturally increases the time you are exposed to the target language, which can only be a good thing. In case you can't recall a word when speaking, you'll still be able to paraphrase, because the dictionary has provided ample training material. Also, trying to translate from one language to the other can harm your ability to output and create a stumbling block whenever a concept doesn't map 1:1 to the target language.
I have to work now. Therefore, just a short answer:
Yes, the "Grammar-translation method" has failed in SLA:
However, this doesn't imply that learners should avoid translations altogether. See, for example: https://www.lucalampariello.com/stop-translating-in-your-head/
The problem for many language learners, esp. at school, is:
- They tend to rely on L1:L2 word correspondences - often without any co- and context information - and that creates a "mess" when they try to write / speak in their L2.
- They don't listen and read enough. Therefore, the "search space" in their L2 is too limited, esp. when it comes to collocations.
From the Anki "thread" that I mentioned:
"Therefore, it's a common complaint among CI practioners (SergeyFM was one of them in this forum a few months ago, and I'm experiencing exactly the same thing in Br. Portuguese right now) that even after reading / listening for more than 1000-2000 h in the respective L2s, we still struggle a lot in speaking (and writing)." (PB)
In other words, only relying on reading / listening in our TLs is "not" sufficient because these are recognition operations that are way too easy - sooner or later.
I don't have a clear answer to that either but I suspect that directionality comes hugely into play. It's MUCH harder to recall active vocabulary when all you have done is passive vocabulary learning (or just stock phrase memorization if you're into that).
I'm speculating/hypothesizing here now:
I suspect it's because passive and active are training two opposite sides of the representational net. You can't expect input to function as output. You need to train an entirely new one and then let them cross-pollinate.
So with that in mind: perhaps.... something like glossika or just flipping round L2 -> L1 cards to L1 -> L2 cards. Or the obvious... practising with natives on zoom.
Maybe it's a good idea to read the LingQ thread "Is 'Anki' A Necessity?" first because you'll find the answer there: https://www.lingq.com/de/community/forum/open-forum/is-anki-a-necessity?post_id=323193
1) It's harder to translate from your first language (L1) into your target language (TL) because there is more mental effort involved, i.e.,
- the active recall of the TL word forms, esp. word groups (= collocations)
- the comparison and correct use of the TL words in a particular co-text (a sequence of words that come before and after)
- resorting to context cues.
This has absolutely nothing to do with "active / passive" because there are no such things as "passive listening", "passive reading" or "passive translations" . The recognition of meaning-form combinations, aka "words / word combinations (collocations / idioms / sentences)" in our TL needs less mental effort than the correct use of these forms. That's all.
So, more appropriate distinctions in this context are:
- recognition / use of individual words, collocations / idioms, etc. - and both (!) are always ongoing operations. Otherwise, nothing happens: no mind, no language.
- focused / (less or no) focused attention = the degree of focused attention (keywords: mono- vs multitasking)
2) When it comes to the context-based recognition of TL forms, reading as a "natural SRS in itself" is probably the best method because you simply get more context information compared to artificial SRS such as Anki & Co.
3) However, there are use cases when artificial SRS make perfect sense. For example:
- Learning specific grammar structures such as verb conjugations or TL cases (e.g., the case system in German, Russian, etc.).
- Low frequency collocations
4) LingQ-to-Anki exports with L1 to TL translations is a nice "intermediary" solution for busy bees
- Use LingQ for your reading (while listening) activities = recognition operations
- Export the LingQs to Anki
- Use the harder L1 to TL translations for the sentences you've previously read on LingQ (the advantage is: LingQ provides context cues and you remember things much better because you've read and / or listen to them before!).
BTW, "intermediary" refers here to the fact that L1-to-TL translations are harder than pure recognition activities, but still easier than using the TL words in free flowing speaking and writing activities (see my reply to @nfera in the thread mentioned above).
Hope that helps,
We did a lot of L1-to-TL translations at university.
Most of us collapsed very fast in those translation courses because just being good at recognizing TL word forms is not good enough for L1-to-TL translations. In short, there's a huge difference between being advanced in an L2 - and having mastered it :-)
I agree in principle with everything you've said Peter but I'd like to quibble with no such thing as active/passive.
Perhaps they're named wrong.
It's pretty evident that going from native language to L2 is harder to recall.
Maybe for the reasons you said, maybe for others (I'm speculating currently about this).
We can call that active recall or we can just call it L1-L2 recall. Regardless... L2-L1 recall and L1-L2 recall seem not to be the same thing.
I hope I'm making sense the direction I'm going here with my speculation.
Yes, usually it's a waste of time to quibble about the "names" for this or that - at least if the phenomena we refer to can be understood.
In this case, it's a bit different because "active / passive" refers to many underlying problems, esp. the sender and receiver model of data transfer between machines where the sender is seen as "active" when it's sending and the receiver is seen "passive" when it's receiving.
It's an excellent model for conceptualizing technical communication, but it's inappropriate for human communication.
And If you ditch this model, the consequences are "huge" because it's a competely new world that opens up, i.e., the world of social "emergence".
And there's not a single phenomenon in the social dimension that is not affected by this change in perspective:
- How do we conceptualize the mind / consciousness?
- How do we conceptualize communication? And how is communication related to action / interaction?
- How do media, esp. language-based media, couple the mind / concsciousness and the social coordination of behavior via communication processes?
- How do we think about interactions, organizations, networks, functional systems like law, politics, science in society and society itself?
- How do we think about "behavioral change"?
But I agree that's not really important in this context. The more important questions for SLA are, for example:
- When is L2->L1 translation useful in artificial SRS (and compared to natural SRS)?
- When is L1->L2 translation useful in artificial SRS (and compared to natural SRS)?
- And if L1->L2 translation is harder, is harder better?
On the one hand, it seems that the "mental resistance" part is important for spaced-repetition processes. In this sense, L1->L2 translations are better than L2->L1 translation.
On the other hand, Florian (@bamboozled) is right: relying on L1->L2 for speaking / writing in our L2s is a recipe for disaster. We know this from the failure of the over-use of the "grammar translation" method.
"So with that in mind: perhaps.... something like glossika or just flipping round L2 -> L1 cards to L1 -> L2 cards. Or the obvious... practising with natives on zoom." (xxdb)
It probably depends on the specific use cases, the language level, and the learner's needs when it is better to rely on reading/listening and when it is better to rely on artificial SRS.
And then there's the question of the "quality" of the SRS cards...
Well, it's complicated... and the cats are not simply black and white, but have various shades of grey - at night :-)
OK gotcha. Shannon model of communication etc.
Target language to known....because you are trying to understand what it is you are reading. You want to be able to see (and hear) the target language word of phrase and be able to understand it. Doing so builds up your "passive vocabulary" or "known words."
The important thing to remember is that you can certainly do this while finishing or preparing for a lesson, or during your down time like standing in line somewhere, etc. Almost all of your vocabularly learning will come from reading and then clicking on the lingq, and not the flashcard.
I will counter with: I use an SPS vocab app for years now and with that I just learn from my native language to the target language! And it works very well.
That's said, to every flashcard expression I add the sentence, where I read/ heard/ said the expression (in the target language).
The other way around I already do with a lot of reading and listening to content.
WANT TO LEARN A NEW LANGUAGE?