Brent Warner 0:00
You’ve probably thought about using AI images to save time. But have you ever considered that generating images can be an active language development experience?

Ixchell Reyes 0:09
In this episode, we will discuss how AI has opened up the possibilities to discover a previously unavailable modality of language production in ESOL.

Welcome to the DIESOL podcast where we focus on Developing Innovation in English as a Second or Other Language.

Brent Warner 0:40
I’m Brent Warner professor of ESL and other things. And I’m, of course always here with Ixchell Reyes, award winning educator in innovation and professional development. Ixchell, how are you doing?

Ixchell Reyes 0:51
Holy moly, well, I’m tired. But I’m excited for this topic, because we just spent two hours going back and forth.

Brent Warner 0:58
Yeah, so. So today’s a little bit different, you know, just a little bit different, I think today might be a little bit of a messy conversation, because there’s a lot of parts here that’s kind of undiscovered or undiscussed in the field. And so we’re talking about image generation, but we’re talking about it on a little bit of a different level today. So let’s jump in and see what we end up with. Okay, so Ixchell, a couple things before we get started. Number one, yes. Okay. We we’ve had a lot of conversations around using AI images to generate an image for your class, right? As you know, hey, I want to put this in my PowerPoint, or I want to put this up on my LMS or something like that. And of course, those things can save you time. Because there you can get quite specific images. And have you done like thing like spending hours just searching for one image that if you have something in mind, and you can’t find the picture that you’re looking for? Yes.

Ixchell Reyes 1:56
And then you give up or you settle for one that isn’t really what you’re looking for?

Brent Warner 2:00
Yeah. So. So we could talk about that in the future. You know, there are ways to develop images and all these things. But we kind of starting to understand that, hey, AI does have that capability to customize our images for exactly what we’re looking for. I also want to acknowledge there are problems with AI images. Have you been paying attention to this Google thing? Oh,

Ixchell Reyes 2:22
absolutely. And I always go back to again, where the beginning, as Eric Roth says, we’re at the beginning of the beginning. So it’s, it’s a rough start, right? Yeah. And it’ll get better. Yeah,

Brent Warner 2:34
so we do have to recognize, like, hey, there are definitely problems, there’s copyright image issues, there’s all sorts of things going on. But we also I personally want to come to the approach of like, Hey, I trust that like, with all things, teachers are not stealing, or trying to steal images, right, that they’re, they’re being proactive that, you know, developers of these programs are trying to figure out ways to be, you know, not get sued, and do these things, you know, as ethically as possible, and all those types of things. So, I’m gonna kind of let we can talk about those conversations in a future episode. And I think there’s a lot to talk about. But today, we’re kind of focused in on you. And I started talking about this and about this idea of using images as actual language production, which is something that hasn’t actually existed before. Right.

Ixchell Reyes 3:28
And so when we started talking about this, I really wanted to know what you meant by that. And so I think it’s worth it’s worth talking about comparing the past versus now. Yeah. And you could because when you say that, searching for an image, or no, not the searching itself, but using an image to within a language activity, that was a passive activity, right? And, yeah, so I’m thinking, Okay, what do you mean by a passive activity? And then you said it beautifully that in the past, an image was what was determining the language that a student might end up using? Right? So the picture is the source of the language. But the student is not actually. I say producing the language. Yeah. And that’s where we’re going to.

Brent Warner 4:28
The language gets a little bit messy around this because it’s time for Yeah, because this doesn’t exist quite yet. Right. And so, anybody out there who is well, well, honestly, we should write and submit to a journal of conversation after after we figure out the language but if someone beats us to it, that’s fine, too. So here’s the thing, right? If you’re in the past when we’ve worked with images in the classroom, traditionally speaking, and there’s variations on this, but a teacher provides an image and a student looks at that image and then they show the their ability to comprehend it with, with what they have, right. So they’re, it’s a kind of a passive experience. So the image comes, they look at the image, they use their language to describe whatever they’re seeing, right? What we’re talking about now with AI image generation is the opposite, right? So now, I’m going to use the language that I have to prompt the bot of whatever it is, and then the bot is going to develop the image of the words that I used. Right?

Ixchell Reyes 5:39
So the language is determining the image, though the language is producing the image and not the other way around.

Brent Warner 5:47
That’s right, the language is producing the image, not the image producing the language. And this has never existed before and could not exist without AI, I don’t think we’re trying to talk super trying to sort this out. And so I think the closest that we’ve ever had to this is, I love this game. So this is like one of the classic language development game activities that teachers do, I think it’s still really powerful and useful. But it’s the one where you sit across from your partner, you put up a cardboard or a piece of paper between the two of you, one of you looks at a picture and tries to describe it to the other person. And then the other person tries to draw right. Now, it’s not the same thing. But it’s as close. It’s a close as the process as we can get to what we’re talking about with now with what we can get with developing images from Ai, because the one person is saying, but again, not by the way, this is still you’re starting with an image or an idea that you’re trying to describe, right. But you’re saying what you’re seeing and someone else is trying to draw the equivalent of that. Now it has to be held inside of your head, and you’re going to try to describe it and then see what the AI is going to produce. But there are differences here, right?

Ixchell Reyes 7:10
Yeah, so if you’ve managed to wrap your head around that, because it took me a while to be able to talk about it in a way that I could conceptualize it. I guess the easiest, what made sense to me is when you talked about the reading versus writing parallel, a student is reading something they’re taking in information, and then they can show you that they understood. But the writing part of it, or the writing aspect is they have to use language to create something. So at the end, there is something they have created. Where with reading, there is nothing really created. Right?

Brent Warner 7:52
Right. Yeah. So you’re either summarizing, or you’re showing that you’ve understood the reading or, you know, you’ve showed what you’ve taken out of it in what what you might learn beyond that, that’s all fine. And that’s the same thing with when you traditionally when a student sees an image in the class, right? They talk they receive that they explain what they’re seeing, they use the language that they have to be able to talk about it. But read just like reading is different from writing. Okay, now now, as I’m writing, I’m putting words together. I’m trying to create something. But the problem for me is that writing in that aspect is still abstract to a student, right? They’re like, I think using the language, I’m not sure, maybe the feedback comes later on parts of it, but not all of it, right? Hopefully, the feedback comes from the teacher, but it might be a week, it might be two weeks, it might not be right away. And so now when we’re talking about this in terms of images, if you’re saying, hey, you know, create an image, if you’re telling, you know, chat, GBT or whatever, create an image. And this is what I want it to look like, then you’re getting that output in the moment. And you’re able to say, Hey, what did my language actually produce? Right? What did the language that I put into that, and we understand that it’s an interpretation of that, but it’s usually pretty good, pretty accurate, you know, again, problems or whatever, but pretty accurate interpretation. And so one of the words that I said, What are the descriptors that I use? What are the verb tenses? What are the whatever? And then what did that produce? And so now we’ve got a whole different level of being able to make and see that what we’re making is, is what we’re intending, right? When we write something with written words, when what we write we hope that what it is, but maybe we don’t have a full enough concept, but as language learners, we all have this visual world that we understand collectively, right? Whether or not we speak the same thing, do we can all see the same images. And so did that image come out in the way that I was in anticipating or that I was hoping or generally right. And so now we have a sense of the language that I used is accurately producing the idea that I was trying to convey without any weight on that.

Ixchell Reyes 10:14
So I think it’s important to mention here that it’s not about that image being the final product, right? That’s not where the language measurement, the development, the measuring of the development is happening. It’s as the student is prompting the bot, the student is making real time decisions about that language. And they’re having to think about, as you said, they’re having to think about what that produced. And as they’re adjusting, it’s like you’re able, the student is able to see that image changing, right? And that is the immediate touch, you’re talking about having immediate feedback for the student. So how much does the language need to be adjusted? How many times do I need to input that? And how how, how does that change that final image, and it’s almost like this gradual process of adjusting language prompts, because that is writing is, is doing another thing that we don’t think about. And that’s, it’s that it is a new way to look at language development in real time. And you should be able to theoretically, look at those prompts that the student is writing and actually pinpoint, like, Oh, my goodness, here is where that language started to shift and where that students started to think about this. And, and you and I talked earlier about this pre show, about how, rather than the teacher being the one giving the scaffolding, as the student adjusts their language, they’re the ones who are laying out that scaffolding for themselves. It’s a it’s a self developed scaffolding.

Brent Warner 11:56
Mm hmm. Yeah. And so this is kind of the idea here is like, we’ve always talked to, you know, a lot of people who are interested in game based learning and gamification, and these types of things are like, Why is it that a student will sit for 12 hours and play you know, I’m gonna go back but like Super Mario Brothers, right? Those play Super Mario Brothers and just like, just to get that one little tiny timing right on the bounce jump to be able to get the top of the flagpole. Right and take it all the way back. But, but we’ll spend a long time doing that, right? And it’s like, and then but with language learning and something that’s gonna value build that value for you. We can’t get it there, right. And so it’s like, what’s going on inside of the game players brain to develop that little skill to get the timing right to get some little little point of perfection, right, just to be able to get the points or whatever. And then that doesn’t happen a lot in, in language learning, right? And what’s the reason for that? Right? And so I’m sitting there looking at that and going. You, you’re boring. It’s like, this language learning experience is not fun, right? In many ways. And so how can we start talking about this stuff? And it’s like, well, because I’m getting that feedback, because I’m able to interact with it immediately, and see what I’m doing and see what I’m making, right. And so, so Ixchell I was talking to you earlier about, you know, I’m just on very low level, I’m trying to learn guitar, and I’m using an app called Yousician. And I can just sit there, my plan is to sit down for 15 minutes, and then all of a sudden, an hour and a half is gone, as I’m just working on this tiny little riff, or I’m working on this, you know, fingering position, or whatever else it is. And so you can get so involved when you’re in control. And when you’re getting that feedback moment to moment, which is something that we’ve never really had a lot of for language development. And I know of course, in the classroom, teachers are giving feedback, but not to not customize individual feedback to 3020, whatever number of students all at once, all at the same time. And so this whole thing of like, where AI is able to do that and make it fun and engaging and interesting for students is Uncharted, right? It has not this not existed before. And I think this is a thing where we’re, we have the arguments, oh, is it good for learning? Is it good for whatever? And it’s like, yes, we have actually there is research out there already, from long years of, you know, edtech integration or whatever. But absolutely right. Students want to do these things. They want something that’s fun and integrated and that shows them their own progress and that challenges them to intrinsically challenges them to say, oh, I can do a little bit better than this.

Ixchell Reyes 14:49
And then the amount of critical thinking that it takes for a student to start putting themselves in what I will say that the bots shoes and going on, and to be able to think about is this same skill that we are using when we’re writing for an audience, right? So that creates that, again, that’s just AI guiding students to make deeper connections to the materials to make deeper connections about the way they’re forming words that they’re forming. And it’s not just a, you know, they’re not typing in one word. Prompt, right? It’s gotta be a phrase or a sentence, and then they’re adding to, and then pretty soon, now you’ve got a paragraph. And now it’s like, well, let me now adjust that paragraph. When when we ask our students to write drafts, you don’t get excited about those little words and adjustments they have to make, by the way.

Brent Warner 15:41
Yeah, no, it’s it’s totally dull. And, and you don’t see, you know, for a student who’s learning another language, you don’t see how rearranging you know, the the clauses in the sentence makes it more powerful or more impactful. But you do see that maybe with an image that you’re like, oh, yeah, that really looks like what I’m thinking about, or that that gets me excited to think about the next part of it, or, Oh, now that it’s, you know, and again, we know that the AI fills in the blanks, right? And so it’s like, it’s going to, it’s not going to do every single pixel as a description, right? But you’re going, but it’s gonna say, hey, it’s putting in these things. And then you might say, Oh, hold on a second. I do want those people in the background, or I want those, you know, instead of you know, this is a science fiction setting that I’m trying to imagine. And I want them flying on jetpacks. But I didn’t think about background people before. So now I’m going to start adjusting those that language as well, you know, so it’s

Ixchell Reyes 16:36
the again, it’s that self scaffolding. And it’s coming as the language is developing, and you’re seeing another version of that image.

Brent Warner 16:46
That’s right. And so when students start putting that into an assignment, so let’s say they they put in, let’s say your final assignment for whatever work you’re doing is pay, put in your, your entire prompting history and then put in the images that generated out of it with each prompt, then like you said, before, we can start pinpointing as teachers and going, Okay, look what you’re learning on your own here. And then look what you’re still missing. Okay, and we can still talk about that, and what what points you can refine and like, oh, I, my guess is that you’re trying to do this, maybe here’s a little language that would help you out with that, right? But there’s, there’s so much cool stuff now. Or we can say, hey, same amount of time that you would be doing something by yourself and not knowing any idea just kind of wandering blindly. Now you’re like, hey, I now I’ve got this guidance, right. And I could still get the feedback. I still get this in that. But it’s not always. It’s not it’s it’s so much more possibility for expansion.

Ixchell Reyes 17:46
Also, I’m thinking Brant, as we’re talking about this. So at the beginning, pre show, I was saying how it’s so important to not focus so much on that final image as the thing. That’s right. Yeah. And I think that as, as students are working with these prompts, and seeing that it, they will come to appreciate what it took to get to that image in the same way that you would feel proud of an essay, because it took you this many drafts and perhaps it’s a little bit more, more connected to you connected with it. So you feel more tied to it. Maybe there’s more of a personal pride. Yeah, well, after that many prompts, man, I would be proud.

Brent Warner 18:32
It’s a little portfolio you’re building actually right as you’re going through, right. And so you can see, then you can really add a glance, see the progress or the changes that you’ve made? Right. And so I think there’s just so much to keep things up as students. I mean, this –

Ixchell Reyes 18:50
Yeah, this is like three dissertation topics right here in this episode.

Brent Warner 18:55
And it’s so exciting because like I said, it has not been discussed, as far as I know, but it hasn’t, it hasn’t been developed as a you know, it’s it’s a new modality of learning, right? It’s a new or a new modality of way of like, approach in CATESOL.

Ixchell Reyes 19:08
Right, right. Fascinating. Absolutely fascinating. Definitely.

Okay, so at the bottom of DIESOL.org. On the homepage, you can find our email subscription list, it’s under newsletter when you scroll on down there. And right there, we have a very clear promise, no spam. And we have stuck to that we we basically haven’t said anything, but we’re gonna start sending out messages. About once a month, I think we’ll see maybe maybe twice depending on how often people want to get updates but some of the information that maybe you’re not able to catch everything or maybe you forget and you want to kind of know what’s going on so we want to send an email on occasion. Nothing too heavy a few things. You know, both of us are now posting more videos on YouTube. to buy and then where you’ve got different social media sites that we’re on, it’s a little bit tricky to keep track of. And so you don’t have to keep track of it, we will keep track of it for you. And we will send out the occasional unobtrusive message and maybe a few links that might be useful or interesting things that we think are possibly worth exploring. So if you’re interested, and if you’d like to join us there, you can get some special bonuses and some other fun stuff. You can go sign up at DIESOL.org, that’s diesol.org and just go drop in your email address at the bottom of the page there. And we would love to have you on the list and join us wherever we are.

Brent Warner 20:42
Okay, so Ixchell we’ll keep this part brief here. But but I’ve written about this stuff in a in a little bit more of a practical application. And we’ve been talking about this in a different way and kind of are now unearthing the pedagogical implications of it, but but I wanted to talk at least about a few of these image generating activities that we’ve talked about in the past, or that I’ve written about and that you’ve written about a little bit too. So brief very briefly, definitely, there’s room for creating images for the teacher to say, Hey, this is going to be something that will help you out. So I’ve done one. Yeah.

Ixchell Reyes 21:21
If you’re interested in mine, mine’s a maybe a quick and dirty beginning type lesson to get you started. I would say,

Brent Warner 21:29
Yeah, it’s great. It’s really interesting, too. Yeah, for sure. So briefly, what what is that?

Ixchell Reyes 21:36
Well, I was interested in generating images for the class, not my students generating the images. But I wanted to have images that would help them to be more interested in of course, I can find images of humans doing things, but I decided to make it into animals. And so for every human character that normally would take, I would have as an image, I changed them into chickens or cats. And because it was funny, it was the, I mean, chickens crossing a sidewalk, but wearing suits. And now that can be you know, that conjures up images of, of, of people in London, I think, is it? Yeah, like the Beatles. So yeah. So. So it’s stuff like that immediately grabbed the attention of my students. And so they’re able to remember later what we had talked about. So that helped to support their language development. So that was kind of where that lesson where that tutorial went. So if you’re interested in that, it’s in our sister blog, and we’ll link it on the show. Show Notes. Yeah, but you had ideas that help. But actually kind of show you what we were talking about. Language Development. Yeah.

Brent Warner 22:52
Yeah, so a couple of -jeez, when was this? I think it was maybe back in December or early January?

Ixchell Reyes 23:00
It’s been a while. Yeah, yours inspired me to try it out. Because I thought, man, my students, I have lower level students. So you don’t have a lot of time. But this is what inspired me to go out there and like, figure out what I could do with it.

Brent Warner 23:13
Yeah, so. So for the TESOL blog. Back in January, I had a post called five ways to use AI generated images in your classroom, and a couple of them are teacher starting, right? So creating images where students have to find the connections between the different images or which one doesn’t belong type of things. But then I had a couple other ones. So the next one, I had an there was idiom adventures, right? Where it’s, you take an idiom that maybe doesn’t make sense to a student, but then the student will then take that idiom, and create an image out of it and make a connection for themselves between the meaning and the actual visualization of it, right. And so, if you’re saying something, it’s like, tickled pink, right? It’s like, well, why does tickled pink mean blah, blah, blah, right? And it’s like, okay, so let me generate some images. And actually, I’m a little, maybe I shouldn’t say that when I’m not sure what would come up. tickled pink, but like I’m on the example I gave on the blog was kicked the bucket, right? And so it started off with like, hey, it’s exactly what you were talking about, too, is this idea of iterating through it, right? And so we say, Hey, first is going to be kicking the bucket. And then it’s like, okay, it’s just someone taking a bucket, but then you kind of say, well, hold on a second, I want it to have death in the background. And then they did it. And it’s like, it was like not wanting to do exactly, but so he said, Hey, now let’s create this image of someone clutching their chest right with death in the background and then but the person was still smiling and so then cut and put a grimace on their face, right and so, so they kept working through the iterations and then being able to see and then we go, Okay, now, as I’m playing with that image, I’m generating that image every single time. Well, now I’m also reinforcing The meaning of the idiom and trying to create an image that connects it to me. And so the students can go through that kind of process. And this is exactly what we’re saying is like, now it’s on them to be able to say, Hey, these are, this is the language I’m using. This is what it’s making. This is how it has meaning to me. And all of that then makes it a much more permanent image for them in their head.

Ixchell Reyes 25:25
I thought that was very interesting. The one with the, with the idiom. Yeah, by the way, fun fact tickled pink has been on several of our episodes (laughter)

Brent Warner 25:38
It’s stuck in our head, you know, you know why that always sticks in my head so much is because I had I had a student run into me randomly and like, saw me and wrote a note and said, “Brent, I’m tickled pink to see you.” And I was like, oh, you know, like, you probably learned that in one of the classes. But like, how often do we really say that? Okay, so. So a couple other ones, grammar grams. So it’s like this one is one where the student can create the image, and then use it to. So we use it to understand a grammar point, right. So the one that I had is one, you can actually draw and create the image first, and then use the language to describe it. But you can say like, Hey, if I know that I want to see two actions going on at the same time, then I might want to be able to use, you know, the present progressive and to be able to say my father, sorry, the past progressive, and then with the simple past, and you say my father was dancing when my mother became upset. And so you’re looking at two different actions happening in the same image, right? But being able to generate that for yourself. That’s one. So you can play with that with all different language points, especially like if you’re getting into mixing them or comparing and contrasting against each other. And not only that, but you can do all sorts of things where students are saying, hey, I want to be able to understand prepositions of location, how things are in relationship to one another, and develop the images that way too. And then the last one that I had here, and Ixchell this is one that I’ve been doing and talking about in workshops, from the beginning, is having people try to describe a famous piece of art and duplicate it through the AI without saying the name of the original piece of art. And I think you play with this a little bit too, right?

Ixchell Reyes 27:26
Yeah, this is cool that you did the silent scream. I think – silent scream?

Brent Warner 27:31
Yeah, just the scream.

Ixchell Reyes 27:33
The scream. I don’t know why I called it the silent scream.

Brent Warner 27:37
Sure, it’s silent. It’s not actually making noise. (laughter) Yeah, so I’ve used that one as a kind of good one. Because it’s a fairly simple image, but classic with with enough details to that you could kind of say, hey, I want to get into parts of these, but I don’t I won’t necessarily miss every point. So like, Hey, do I say like orange oil, swirly images or whatever, right? And so depending, but you can play around with it with your students and kind of see what happens there. But But it’s great, because then again, same thing is that they can challenge each other. If you’re in class together and say, who’s gonna get the closest image? Right? Who’s gonna get it closest? Or who’s gonna get the one that’s most interesting? Oh, yeah, they can vote and you can do a little little competitions with each other. But again, they’re gonna keep playing with it and say, Oh, why did why did you get that language? Right? But I didn’t get it and yours looks closer, or mine looks. Mine looks different. What what did I miss, you know, and so again, it’s about that processing of their own language and seeing that output and then being able to interact with it and make changes based on the language that they thought they were using.

Ixchell Reyes 28:37
Right. Very, very cool. Yeah.

Brent Warner 28:42
So there’s a lot to look at here. I think there’s far longer discussions going on in the future here. But uh, but anyways, there’s plenty. We’re posting about this on AI and ESL sometimes. And check the show notes for more information or more examples because we got a few other ones that we’ve played with too.

Ixchell Reyes 29:00
All right, it is time for our fun finds. And today I have Trader Joe’s ginger juice powder. juice powder, ginger juice you like none of those words go together right, that little kick. After you sip on some water you where you’ve mixed the powder and the health benefits of ginger then that’s your mix.

Brent Warner 29:27
Trader Joe’s makes all these great little things like that. Wonderful. Okay, so ginger juice powder, you can you can mix it into your smoothie or whatever else to write so awesome. Mine is an Instagram account this time. I never I don’t think there’s a right way to say this but the spelling is MYLLCK.

Milk!

My lick? (laughter). Something like that. Anyways, it’s just a great music account and it’s this father son team, I guess I think they’re in like Singapore or something like that, but they’re both just like super into music. And then what they do is they show you Yeah, they do these like little trends, they’ll be like, okay, great Brazilian albums that came out, you know, from the 1980s or something and it’s like, okay, and then they just give you five or six of them. And I have just discovered so much cool music because of his like world music or different styles of things. And so and they’re, they’re both just so happy looking to they’re just like so stoked to be sharing music with everybody. So it’s a great account if you’re interested and want to kind of discover some different music in a new way. MYLLCK

Ixchell Reyes 30:41
You can find us on Threads or on Facebook, we are at DIESOLpod on both platforms, right?

Brent Warner 30:47
You can find me on the socials at @BrentGWarner and

Ixchell Reyes 30:51
me at @Ixy_pixy, that’s I x y underscore p i x y

Brent Warner 30:57
And for our mystery language global I think we’re gonna have to find a better a better terminology for this in the future but mystery language global phrase let’s let’s have a listen.

Mystery Phrase 31:07
(XXXX)

Ixchell Reyes 31:12
If you can figure out what this means, let us know on social media tag us at DIESOLpod.

Brent Warner 31:19
Thank you so much for listening to the DIESOL podcast.

Ixchell Reyes 31:22
Thank you.

Brent Warner 31:23
Take care everybody. Bye