I had a similar idea for a custom gpt. Start with a sentence and then swap components. I ate pizza. He ate pizza. He made pizza. He made a cake. **But then add a competent to the sentence. He made a cake yesterday. They made a cake yesterday. They bought a cake yesterday. They bought a car yesterday. They bought a car last week. And just add on to the sentence after swapping out every component. “They bought a new car last week.” Having the GPT give prompts while in English while the user tries to type it in their L2. And just do that to starting with a simple sentence and getting more complex until it gets to the point where just before it should be 2 sentences.
Oh....this explains why the CI-plus-sentence-mining approach works so well for me. My brain just, uh, does this when I do sentence mining and learn new words. But! I can definitely be more intentional about using it. I had this...strange thought that I couldn't do this (at least previously for Chinese) because I wasn't sure the results would be correct? Idk. I can certainly do this for German now - I mean I can speak fine already, but there's definitely a few structures I could just do with more specific practice in so they become as automatic as the other structures that have somehow sorted themselves out.
I like this approach. Although I am learning through understanding (meaning to learn something I have to understand how and why something works) most people learn using patterns, and this method may be very useful to them. I also use this approach in the beginning, when sentence structure is pretty given and doesn't change while swapping words: like this is a dog, this is a house etc. (not all languages, though, Japanese has two verbs to be: for animate and inanimate objects, but it's easy to remember), and this is why I often advise people to go through phrase books for fun, because they use fixed phrases. But while at it, remembering bits of grammar, like adjective-noun gender accordance in French or Italian, or the role of imparfait as a "background" for action (parfait) in French are pretty useful. You don't have to learn the whole grammar, but sometimes remembering a rule is faster than remembering the whole bunch of sentences you just practiced. I love how you called pure CI a "hope-based methodology" - couldn't be more accurate!
@@Evildea I guess middle ground is the key. Some languages really are difficult grammar wise and you won't be always able to tell the rule from the sentence. For example: Daję coś dwóm dziewczynkom. (I am giving something to two girls). Daję coś dwóm osobom. (I am giving something to two people) - so far so good, the rule is forming, right? Okay - Daję coś dwojgu ludziom. (I am giving something to two people.) If you don't know, that osoba is feminine and człowiek is masculine and that its plural form is ludzie you will be lost. Eventually you will figure it out, but it's gonna take a looooot of time. It's easier to remember, that osoba declines similarily to dzewczynka and człowiek changes to ludzie in plural and has a masculine form. Polish is not the only language with a complex grammar and it will take ages to figure out all sentence patterns, similarily in Arabic, Hungarian, Finnish, Danish... Rules help to navigate languages like this a lot and remembering them is much much faster to learn, than go through gazillion of version of the same sentence.
@ Definitely, some languages you're just gonna be forced to learn certain aspects of grammar early on while with other aspects you really don't need to know the grammar right away and you'll make faster progress with patterns..
In college we did drills sometimes, and I really liked them. The first was in Hebrew. We sat in a large circle and verbally passed the pattern around the room. Every so often the professor would call out a new word or some change up to the pattern. We also did this in Russian. It was pretty similar, but this time English sentences were verbally passed out to translate, but each time there was a small variation to the pattern.
I'm used to calling these "substitution drills." I agree, they are powerful. We did this a lot of this in beginning Spanish class, junior high in Florida in 1968. (You can, btw, do this on the fly, without any aids: pausing a video or looking up from your book to play with a sentence.)
Ah yes, I’ve heard this term once before. I wasn’t sure though how closely it reflects what I do so wasn’t sure if it was the right term for what I do.
@@Evildea We used it to describe the following: Students are given a sentence-cd be on paper, or spoken. They are then asked to substitute other words (usually from a list, or with an oral prompt) into the place of one specific word of the sentence.
Read Cross agrees with this approach. Works well with group learning for building known words. The "mass input" style needs lots of materials, not suitable for an anywhere, low resource reality.
Well, what you call the template based approach IS the grammar approach. You learn a sentence and you apply the grammar rules to change the specific part that you are practicing and you do that until that process becomes automatic. :D
And to be more pendatic, what people call grammar rules are actually grammar patterns! It is just a description of the patterns people follow to convey specific meanings.
A grammar approach would break it down into each individual piece and learn the rules that govern it. A template approach doesn’t care about the rules. Only what can and can’t be swapped out. At no point does a template require you to understand why the structure is the way it is. You just learn the structure.
How is this different from Grammar Translation method? Putting it in an interactive computer program seems like a nice idea. When I read lists of such sentences in books, my brain seems to go numb. Maybe it needs to be 'gamified' a bit although I detest that word as of late.
I'm confused by the idea that pausing to think through some kind of grammar calculus during conversation is an actual methodology, because I feel like I only ever hear about this from people saying, "No, do this other thing instead." lol When I think of grammar study, I'm pretty much thinking of grammar as just the patterns of the language. I just have a sloppier "run through it once or twice and try to use it in conversation to see how it goes" approach.
Thank you for the inspiration! Can I apply the template based approach to languages like Japanese where there are lots of conjugations without learning any conjugation rules? Or do I need to learn the conjugation rules simultaneously to speed up the process?
Hi, speaker of both Japanese and Chinese here. I would say that the template method is much easier to apply to Chinese than to Japanese. For Japanese, I would recommend at least having a basic knowledge of the needed conjugations (since there are irregulars) and then using the template method to internalise these conjugations, so that they become second nature.
You can. I actually did this for Latin by mining sentences from Latin text books. Latin is heavy in conjugation rules. You will naturally through that process of acquiring these templates get a surface understanding of the grammar. However, you won't need to manually learn or apply it. Once the templates are created and you start the exercises as @EminjoenJapanio says, you'll start to internalise the patterns.
@@Evildea @EminjoenJapanio Thank you for sharing the valuable experience! My previous thought was that once I learn the most conjugation rules and then apply and practice them, I can cope with the most conjugations even with the verbs I would encounter for the first time. Because I thought without learning the conjugations deeply, I have to encounter all the verbs I need with every conjugation patterns in some way. But you mean doing the sentence mining and practice them over and over again is still more efficient and effective than applying and practicing the conjugation rules? I'm just wondering this part. Thank you!
Evildea: I'm gonna try keep this super short
7:32 minutes later...
Evildea: ...
Discord: discord.gg/tPbQTtU2Zt
I like the specificity. So many advice videos on YT are just generalities with little meaning.
Thanks!
That is true..almost no language learning videos actually talk about techniques.
I had a similar idea for a custom gpt. Start with a sentence and then swap components.
I ate pizza.
He ate pizza.
He made pizza. He made a cake.
**But then add a competent to the sentence.
He made a cake yesterday.
They made a cake yesterday.
They bought a cake yesterday.
They bought a car yesterday.
They bought a car last week.
And just add on to the sentence after swapping out every component.
“They bought a new car last week.”
Having the GPT give prompts while in English while the user tries to type it in their L2. And just do that to starting with a simple sentence and getting more complex until it gets to the point where just before it should be 2 sentences.
Yeap, that would also work very well and it's something I've done.
Oh....this explains why the CI-plus-sentence-mining approach works so well for me. My brain just, uh, does this when I do sentence mining and learn new words.
But! I can definitely be more intentional about using it. I had this...strange thought that I couldn't do this (at least previously for Chinese) because I wasn't sure the results would be correct? Idk.
I can certainly do this for German now - I mean I can speak fine already, but there's definitely a few structures I could just do with more specific practice in so they become as automatic as the other structures that have somehow sorted themselves out.
Yeah, that's the perfect way to do it.
I like this approach. Although I am learning through understanding (meaning to learn something I have to understand how and why something works) most people learn using patterns, and this method may be very useful to them. I also use this approach in the beginning, when sentence structure is pretty given and doesn't change while swapping words: like this is a dog, this is a house etc. (not all languages, though, Japanese has two verbs to be: for animate and inanimate objects, but it's easy to remember), and this is why I often advise people to go through phrase books for fun, because they use fixed phrases.
But while at it, remembering bits of grammar, like adjective-noun gender accordance in French or Italian, or the role of imparfait as a "background" for action (parfait) in French are pretty useful. You don't have to learn the whole grammar, but sometimes remembering a rule is faster than remembering the whole bunch of sentences you just practiced.
I love how you called pure CI a "hope-based methodology" - couldn't be more accurate!
Yeap, like all things, people will figure out the sweat spot for themselves. Some like little grammar, some love it a lot.
@@Evildea I guess middle ground is the key. Some languages really are difficult grammar wise and you won't be always able to tell the rule from the sentence. For example: Daję coś dwóm dziewczynkom. (I am giving something to two girls). Daję coś dwóm osobom. (I am giving something to two people) - so far so good, the rule is forming, right? Okay - Daję coś dwojgu ludziom. (I am giving something to two people.) If you don't know, that osoba is feminine and człowiek is masculine and that its plural form is ludzie you will be lost. Eventually you will figure it out, but it's gonna take a looooot of time. It's easier to remember, that osoba declines similarily to dzewczynka and człowiek changes to ludzie in plural and has a masculine form.
Polish is not the only language with a complex grammar and it will take ages to figure out all sentence patterns, similarily in Arabic, Hungarian, Finnish, Danish... Rules help to navigate languages like this a lot and remembering them is much much faster to learn, than go through gazillion of version of the same sentence.
@ Definitely, some languages you're just gonna be forced to learn certain aspects of grammar early on while with other aspects you really don't need to know the grammar right away and you'll make faster progress with patterns..
@@Evildea Exactly, which is why: learn ABOUT the language you want to learn and be sure you want to plunge in 😁
In college we did drills sometimes, and I really liked them. The first was in Hebrew. We sat in a large circle and verbally passed the pattern around the room. Every so often the professor would call out a new word or some change up to the pattern. We also did this in Russian. It was pretty similar, but this time English sentences were verbally passed out to translate, but each time there was a small variation to the pattern.
Very interesting!
I'm used to calling these "substitution drills." I agree, they are powerful. We did this a lot of this in beginning Spanish class, junior high in Florida in 1968. (You can, btw, do this on the fly, without any aids: pausing a video or looking up from your book to play with a sentence.)
Ah yes, I’ve heard this term once before. I wasn’t sure though how closely it reflects what I do so wasn’t sure if it was the right term for what I do.
@@Evildea We used it to describe the following: Students are given a sentence-cd be on paper, or spoken. They are then asked to substitute other words (usually from a list, or with an oral prompt) into the place of one specific word of the sentence.
Looks like a solid method 👍👍👍
Thanks!
Read Cross agrees with this approach. Works well with group learning for building known words. The "mass input" style needs lots of materials, not suitable for an anywhere, low resource reality.
True that
With Read Cross we start with 16 nouns to cover as wide a range of things as possible. Ask yourself, which 16 would you choose?
Well, what you call the template based approach IS the grammar approach. You learn a sentence and you apply the grammar rules to change the specific part that you are practicing and you do that until that process becomes automatic. :D
And to be more pendatic, what people call grammar rules are actually grammar patterns! It is just a description of the patterns people follow to convey specific meanings.
A grammar approach would break it down into each individual piece and learn the rules that govern it. A template approach doesn’t care about the rules. Only what can and can’t be swapped out.
At no point does a template require you to understand why the structure is the way it is. You just learn the structure.
@@maxjahnke true.
@@Evildea "You just learn the structure." - you mean - cram the structure?
Yes
Good.
Such drills are smart and yes, will help you learn faster than pure CI
Yeap, combining both is even better :D
Thanks for explaining this in detail!
No problem :)
This is very useful 👍
That's good to hear :D
I want to go to: the ice cream shop, the park, the movies, the beach, the skate park, the pub. Ha, could have shown us with a spreadsheet 😎👍
True, but I was coding a system so figured I'd just show you with code :)
Sounds similar to Tony Marsh language matrix method using statements, questions and answers.
Good to hear!
could you release this code and provide instructions on it
its would be like giving the cure to cancer
Once I've written it more I'll release it. It's only 2 days old at the moment.
That is actually an idea I had came up with for myself. So it's cool to see I'm nkt the only genius 😎 😂
Many big brain peps out there :P
How is this different from Grammar Translation method? Putting it in an interactive computer program seems like a nice idea. When I read lists of such sentences in books, my brain seems to go numb. Maybe it needs to be 'gamified' a bit although I detest that word as of late.
I'm confused by the idea that pausing to think through some kind of grammar calculus during conversation is an actual methodology, because I feel like I only ever hear about this from people saying, "No, do this other thing instead." lol
When I think of grammar study, I'm pretty much thinking of grammar as just the patterns of the language.
I just have a sloppier "run through it once or twice and try to use it in conversation to see how it goes" approach.
People actually do pause and think through grammar. I've had conversations with people who have tried this lol.
@@Evildea I hope they push through it pretty quickly. Sounds rough.
More programming with languages please...live programming with commentary.
Thank you for the inspiration! Can I apply the template based approach to languages like Japanese where there are lots of conjugations without learning any conjugation rules? Or do I need to learn the conjugation rules simultaneously to speed up the process?
Hi, speaker of both Japanese and Chinese here. I would say that the template method is much easier to apply to Chinese than to Japanese. For Japanese, I would recommend at least having a basic knowledge of the needed conjugations (since there are irregulars) and then using the template method to internalise these conjugations, so that they become second nature.
You can. I actually did this for Latin by mining sentences from Latin text books. Latin is heavy in conjugation rules. You will naturally through that process of acquiring these templates get a surface understanding of the grammar. However, you won't need to manually learn or apply it. Once the templates are created and you start the exercises as @EminjoenJapanio says, you'll start to internalise the patterns.
@@Evildea @EminjoenJapanio Thank you for sharing the valuable experience! My previous thought was that once I learn the most conjugation rules and then apply and practice them, I can cope with the most conjugations even with the verbs I would encounter for the first time. Because I thought without learning the conjugations deeply, I have to encounter all the verbs I need with every conjugation patterns in some way.
But you mean doing the sentence mining and practice them over and over again is still more efficient and effective than applying and practicing the conjugation rules? I'm just wondering this part. Thank you!
Are you writing code to generate this? If so can you explain what programming language you're using and the process?
It is written in C++. I'm just writing it off the top of my head.
You can use these same techniques using AI--it's a great learning and learning reinforcement tool. Try it!
Yeap, I've actually spoken about this before.