Pratique du Shadowing: The AI Tool You Actually Need for Ancient Greek (NotebookLM) - Apprendre l'anglais à l'oral avec la vidéo

I want to share something with you today which I am exceedingly excited about.

And it starts with the fact that there are virtually no Greek speakers alive today who can speak ancient Greek fluently.

There's plenty of people who can speak Latin.

There's not many people.

There's only a handful who can speak Greek fluently.

And the problem here is not with the difficulty of the language.

And we know that because for a thousand years, there were generations of people doesn't matter if they were old or young, doesn't matter if they were clever or dumb,

doesn't matter if they were educated or uneducated, who spoke Greek fluently.

We know that because it was a living language.

So what is the problem today?

It's not the difficulty, it's actually because we don't have an immersive environment the same way that we have for modern languages.

Which means that if you want to learn to speak ancient Greek, you have to reconstruct that immersive environment, which you can do by listening,

obviously, and you can do that by reading.

The problem is that most people don't read enough

but also we have a skewed corpus

which means we have a lot of literary works we have history

and we have philosophy and that kind of thing but very little

which is related to the everyday interactions which is the kind of thing

that we need to be able to speak so

if you want to learn how to say things like you know i get dressed

or i go for a shower you're going to have to

do a lot of reading just to stumble across those phrases which you need in order to be able to speak fluently.

And the other thing that you need is an environment where speaking is possible, right?

You need friends to communicate with and that kind of thing.

So here's the problem, okay?

In order to be a speaker of ancient Greek, you need three things in today's world.

First of all, you need to be highly motivated because you have to reconstruct this entire immersive environment

yourself and that requires motivation.

You also have to be reasonably clever because the environment

which you construct is not going to be of the same quality as going to live in ancient Greek

and hearing Greek around you all of the time and having to use Greek in your daily conversations as well.

And the third thing

that you need is the opportunity to be around other speakers of ancient Greek and ideally have an ancient Greek speaking teacher.

That's a problem for most of us as well

because not all of us are at leisure to go and study in Jerusalem or to pay a private tutor

premium for their ability to speak ancient Greek.

Now there is a solution.

I want you to imagine a situation where at any point in your day you can get out your AI

and you can ask it questions about your daily life.

For example, let's say I'm in the kitchen and I'm doing the dishes.

I ask AI, AI, how do I say I scrubbed the pot in ancient Greek?

And it gives you an answer.

Okay, now you can already ask AI that.

The problem is that AI is pretty terrible at ancient languages at the Okay, Latin it's okay at because it's being trained on a massive corpus of Latin.

Greek it's much worse at, okay?

It makes mistakes all the time, okay?

It's very unidiomatic and even

if it does get it right you don't actually have a source

which can tell you exactly where that information came from in the first place

which is very handy if you're doing academic work let's say

or you're trying to help students and you want to make sure

that the idiom

which you're looking for is actually attested in greek somewhere now

ai will get good you know it will take another five years

and it will get good but the hack

that I'm going to show you today is going to put you ahead of the AI curve by five years.

It's going to give you what you want right now.

You can ask AI anything which is related to your daily life right now.

So you might be in the gym and you ask AI how do I say I lift weights?

How do I say you know I do a squat or fry my eggs?

Or what about interactions?

Things like things that we say all the time like you know good to see you you know I'll catch you later.

Now, these phrases will exist, you know, if you go and read an actual Greek text, like right now I'm reading Xenophers' Anathesis.

But it's going to take you a lot of reading to stumble across these phrases, simply because, like I said before,

the corpus is skewed and it doesn't have the kind of interactive, everyday phrases which you need in order to become a speaker of the ancient languages.

Now, there are two solutions which I have tried out.

The first solution is the solution which is next to me, which is this book called Hopparos.

And it says here that it's based on the New Testament, Marcus Aurelius, Josephus, and 22 all-up Koine texts.

And the title here is A Thematic Guide to Ancient Greek Vocabulary and Set Phrases, which is very handy because it means I can go and look up a topic.

So if I open my book, it says here to swim or to dive.

So if I want to know how to go for a swim, I look up this topic, and it has all the vocabulary, including a bunch of set phrases, which is going to help me if I want to speak ancient Greek.

So I found this book very helpful, and you will too if you buy this book.

The other solution that I've come up with, and I'm going to share my screen with you, so let me make myself a little bit smaller here,

is that I started collecting phrases like this, as you can see on my screen.

So I was reading, like I said, I'm reading at the moment, the anabasis and I went through and I tried to collect as many idiomatic phrases

which I could find and then put them into a Google Sheets like this.

So for example, food and drink is sita kai pota

and then I would have all these tags right and these descriptors

so that later on I can search for them and then hopefully find the phrases which I need.

And that would have worked really well, it will work really well eventually.

The problem with this however is that it is, let me make myself just a tiny bit bigger here.

The problem with this is that this is very time intensive.

Okay, if I spend an entire year working at this consistently, I might get to, you know, 1,500 expressions, which is good, right?

I'll be able to search that and it would help me.

But the hack that I'm going to show you today takes, you know, 10 minutes of work once you know how to do it.

And you've got basically an infinite mine of sentences like exactly as I've got here.

And it takes virtually no work to look them up either.

You just ask your AI.

So we're going to use Notebook for this.

And this is not a tutorial.

If you want me to make a tutorial on this, let me know in the chat and I will do that for you.

But I'm going to show you the finished result.

So I've created this notebook.

which is on Xenophon's Anabasis.

I'm just going to go here and delete the chat history so that we can start afresh, which is basically, this is a Google AI.

And I have uploaded the entire Anabasis here.

And the instructions that this notebook essentially has is that for any question I ask it, it has to use my sources, the sources that I have uploaded,

rather than its general training, which is great because the general training of AI is not that great.

AI was made by software developers, by definition, and software developers are going to make AI very good at developing software first,

100

but they're not classicists and they don't really care about how good AI does in ancient Greek, which is why they haven't trained it on much ancient Greek data.

101

But I can force it to use exactly the text

102

that I care about and make it pinpoint accurate for every answer it gives me about the ancient Greek language.

103

So let's see how this works.

104

we've got here the source anabasis

105

and I'm planning on making an entire compilation right a bit like the Faris where I've got 20 odd authors

106

and that I care about and that have diction

107

which I consider normative for the way

108

that I want to be able to speak I'm going to upload them into a single notebook

109

and then every time I ask it a question it will

110

search through every single one of those authors to find the phrases

111

that I'm looking for so at the moment I have the anabasis here

112

and I'm just going to show you again what I put in my language islands

113

because I have a few expressions here which we're going to test out

114

so I found tez deiles for in the afternoon okay

115

and there's also some other phrases related to the breaking of the day

116

so epedan hemera genetai when the day breaks so let's test this

117

and see how good this works so I'm going to ask it

118

how would Xenophon say in the afternoon as a time phrase and the expression when the day breaks

119

and I will let it think

120

and it's going to take a little bit longer than a normal AI

121

because it's actually searching the text that I wanted to search

122

which means there is going to be no hallucination whatsoever so let's give it a moment All right,

123

dayless or test dayless.

124

Now isn't that interesting?

125

That's exactly what I had and it's got sources as well, which means we can click on it.

126

Here we go.

127

And if you look at this it's got the section highlighted

128

which i care about

129

and somewhere in here it's going to have day list there we go day list this time without the article

130

but this is the beauty about this every single answer it gives you is rooted in actual greek text

131

so if you're doing academic work this means

132

that you can actually cite ai

133

which is that's crazy right i mean it's got some other expressions here

134

which i didn't even know about

135

and it would have taken me i guess a long time to actually find this in the greek text

136

that i've been reading and now the other expression

137

which we were looking for was when the day broke

138

and it doesn't have the expression

139

which i have okay it's got um hemera who pep pines are using a different verb actually now

140

that doesn't mean this ai is wrong in fact i'm more likely to trust this ai than myself

141

because it's looked through the entire analysis and found the expressions which are the most common.

142

So obviously my language island where I had Epidone Hemera Genetae, but obviously correct because it occurs, right?

143

I've got a, somewhere here, I've got a citation for that as well.

144

It's obviously correct, but it might not be the most common and that exposes, I guess, another fallacy of doing this manually.

145

So I will eventually gather all the texts that I care about

146

and put them into one notebook where I can search through, you know, all of Attic literature

147

or maybe I'll have two separate notebooks one for Attic one for Koine

148

but there's going to be no Homer in there there's going

149

to be no Ionic there's going to be no poetry it's only going to be the language

150

that I consider normative for the way

151

that I want to be able to speak now you may have noticed before

152

that I've done a very similar thing for the New Testament

153

and I hope to add the Septuagint to this very soon as well

154

because I teach a class where I try to teach the New Testament vocabulary in an immersive interactive way

155

and for that I want to make sure

156

that all the expressions I use actually occur either in the Septuagint or in the New Testament

157

and so this is going to be very helpful for me from now on.

158

Now the New Testament's a little bit of a different kettle of fish than the entire TLG,

159

the entire corpus of Greek literature because AI has seen the New Testament a lot of times, right?

160

It gets quoted all of the time on the internet

161

which means it's seen it in its training data over

162

and over again

163

and it's it's able actually already as is in a normal chat to extract data from the new testament okay

164

but it isn't able to do

165

that with you know a vast corpus of classical literature unless we do what i'm showing you right now

166

which is to designate the text

167

that you want it to search through now like i said

168

if you want me to show you a tutorial on how

169

exactly to set this up let me know in the in the comments now Now, for those of you who use Enki, I made this other video in

170

which I show you exactly how you can make a 100% customized Enki deck using AI within a matter of minutes, right?

171

Without having to type out all of the cards manually.

172

So make sure you check that out.

Qu'est-ce que la technique du Shadowing ?

Le Shadowing est une technique d'apprentissage des langues fondée sur la science, développée à l'origine pour la formation des interprètes professionnels. Le principe est simple mais puissant : vous écoutez de l'anglais natif et le répétez immédiatement à voix haute — comme une ombre suivant le locuteur avec un décalage de 1 à 2 secondes. Les recherches montrent une amélioration significative de la précision de la prononciation, de l'intonation, du rythme, des liaisons, de la compréhension orale et de la fluidité.

Pratique du Shadowing: The AI Tool You Actually Need for Ancient Greek (NotebookLM) - Apprendre l'anglais à l'oral avec la vidéo

Context & Background

Top 5 Phrases for Daily Communication

Step-by-step Shadowing Guide

Qu'est-ce que la technique du Shadowing ?