Getting the most out of LWT

[neuroascetic]

Adrianslont,

I used to not do the flashcards from Readlang, but mostly because it's easy to create so many and it's hard to keep up with the reviews. I've found that Readlang's feature where when you are in a text, it will highlight words that have an associated flash card that is due (according to the srs algorithm), and you can do a repetition right then and there. It is pleasurable to see the number of highlighted words decrease as you read further along, suggesting that your comprehension is getting better.

However, I've recently tried to keep up with the flash cards. My rule is that I try to do 100 flashcards per day and if I'm reading a book, I'll get caught up on all the words in that book before moving on to the next chapter (you need the subscription for this feature). The issue is that I came up with this rule while I was about 7 chapters into the second Harry Potter book and had 1500 words to review. It's taken me about a month (maybe more, I'm not sure) to drill these words. So, I actually haven't read another chapter since I started this, so take that for what it's worth.

I'll report, though, what I'm finding. I find it valuable to keep the flashcards in Readlang. I don't like having my cards in multiple places (I'm currently doing repetitions in four systems right now) and assume that the Anki and SuperMemo algorithms are much better. One reason I keep the cards in Readlang is that it automatically sorts the cards by word frequency. Though I'm sure it's not perfect, it works well enough. An important feature is that a card can hold multiple contexts (one gets shown to you randomly each time you see a card). So, I can see the different ways that a word has been used in a text. That can be useful when I get a card and can't quite get the meaning, but if I check a couple of other contexts showing how it has been used, I can remember the meaning, without checking the translation.

There's a rule that I use when reviewing Readlang flashcards that I find to be helpful. When I read the example sentence, I don't translate anything in my head except for the key word. Sometimes this means that my understanding of the sentence is poor because there are multiple words that I don't know. However, those other words have their own flash cards. What happens is that as I come back to these sentences over and over, my semantic understanding of the sentence improves, and these sentences make more sense and I don't have to translate. Along the same lines, I try to hit the button to see the L1 answer at that moment where I can remember the meaning of the L2 word, but haven't yet retrieved the translation from my memory.

So what's the upshot of all this. Over the past month of doing this, it's improved my comprehension quite considerably. More so than if I spent that time reading the rest of the book, I'm pretty sure. I can tell that there's been an improvement in my listening comprehension in that I'm able to pick out many more words and work out the meaning without translating as much as I used to.

It also improved my confidence enough that after three years of wanting to go to a local speaking group, but being too nervous, I finally went to my first one last week and it went much better than I expected. So much so that I went to another one a few days later. This feels significant.

A couple improvements I would like to see. I wish Readlang had a way for to use monolingual definitions instead of translations for flashcard answers, or at least, allow you to score a card without seeing the translation if you know it. I wish there was an easy way to filter the cards by frequency (i.e., if a word is a certain level of obscurity, you don't see the card). Readlang also isn't good about separating cards that you have started and cards that are new which makes the number of cards to review seem more oppressive. But, otherwise, it's been a great tool to use for building vocabulary.

[neuroascetic]

Also, Readlang could really use a "leech" feature.

[Josquin]

Although I use LWT on an almost daily basis to read Spanish, French, and mostly, Ancient Greek, and I have used it in the past to read German,

MorkTheFiddle, I'd be interested in how you go about working with Ancient Greek inflections in LWT. Irish doesn't have a lot of conjugations and declensions, but it does have initial mutations, so I have to enter a word in several forms before LWT really "knows" it. On the one hand, this makes you learn the inflections by heart, on the other hand, it's a lot of work. But, as I said, this is manageable for a language like Irish.

However, I now want to use LWT for Russian and, maybe, later for Ancient Greek, too. Both of these languages have plenty of verbal and nominal forms all of which you would have to enter into LWT every time you meet an unknown word in a new form. How do you handle this problem? Do you only enter unknown words once or every time you meet them in a different form? Is there some kind of work-around for this problem? Can one "teach" LWT to recognize inflected forms?

Any advice is once again appreciated!

[Adrianslont]

Also, Readlang could really use a "leech" feature.

Hi Neuroascetic
Thanks for the replies to my questions. I know what you mean about being too easy to create more cards than you can keep up with in Readlang. I compound the problem when I add words just because I want to hear them pronounced! I overcome the problem to some extent by frequently deleting cards when I do reviews - if it’s actually a word that keeps repeating in the story I delete it. If I feel I don’t need to keep seeing it, I delete it. If the sample sentence doesn’t seem very good quality, I delete it. If I decide the word annoys me, I delete it. This helps keep the number of cards down a little but I still have too many. You can be your own leech algorithm!

I am looking forward to seeing where my reading is by Christmas. I started Readlang about 35 days ago now and I really hadn’t done a lot of reading before that - most of my learning was listening in the car or anki at home plus some reading in Indonesian. Now I am severely distracted by reading in French. I think Readlang would be even more helpful with Indo than French because it doesn’t have many English cognates.

[MorkTheFiddle]

Although I use LWT on an almost daily basis to read Spanish, French, and mostly, Ancient Greek, and I have used it in the past to read German,

MorkTheFiddle, I'd be interested in how you go about working with Ancient Greek inflections in LWT. Irish doesn't have a lot of conjugations and declensions, but it does have initial mutations, so I have to enter a word in several forms before LWT really "knows" it. On the one hand, this makes you learn the inflections by heart, on the other hand, it's a lot of work. But, as I said, this is manageable for a language like Irish.

However, I now want to use LWT for Russian and, maybe, later for Ancient Greek, too. Both of these languages have plenty of verbal and nominal forms all of which you would have to enter into LWT every time you meet an unknown word in a new form. How do you handle this problem? Do you only enter unknown words once or every time you meet them in a different form? Is there some kind of work-around for this problem? Can one "teach" LWT to recognize inflected forms?

Any advice is once again appreciated!

I enter each and every form of a noun, adjective or verb as I encounter it. An entry would look like this:

τάξω / * thou marshalled. aor 2s of τάσσω

The good news is that eventually you learn enough so that you can leave some things out. For example, if I ever can remember that the aorist of τάσσω changes the σσ to ξ, then I could just write

τάξω / * thou marshalled. aor 2s

, and as time goes by, slowly drop other parts of the definition:

τάξω / * thou marshalled

, and so on.
I was much too far along before I realized that I could enter the main form of the word in the 'tag' box.

Learning all the inflections helps, but learning all the verbal inflections at the very beginning would be a formidable task. I do recommend learning the first and second declensions of nouns. Also, don't overlook the aorist, because the aorist stem is often rather different than the present stem. Eventually you will read enough to begin to recognize some structures without much study. And even realize precise knowledge is not always necessary. A word in such and such a place in a group of such and such words will have to be in one and only one case. That's what I have found, anyway. Nevertheless, I do from time to time work formally on accidence that I don't know yet.

To answer your final question, LWT cannot to my knowledge be taught to recognize inflected forms.

Hope this helps. Good luck.

[Josquin]

Thank you very much! This is most helpful.