Romanistan

Rowena Marin on Google Translate and the Romani Language

Jezmina Von Thiele and Paulina Stevens Season 4 Episode 17

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 37:06

This episode explores the inclusion of the Romani language on Google Translate, featuring Rowena Marin's insights on the intersection of technology and culture. We discuss the journey from grassroots initiatives to digital recognition, the complexities of language preservation, and the opportunities technology can create for the Romani community.

• The journey of Rowena Marin from Romania to Google
• Discussion on the grassroots campaign for Romani language inclusion
• Detailed process for adding the language to Google Translate
• Concerns surrounding AI and authenticity in language translation
• Evolution and representation of different Romani dialects in tech
• Opportunities for promoting Romani literature through digital platforms

Resources: Roma Education Fund 

The Internet learns Romani 

Listen to our first interview with Rowena Marin about her book Who Am I in the World? 

Thank you for listening to Romanistan podcast.

You can find us on Instagram, TikTok, and Facebook @romanistanpodcast, and on Twitter @romanistanpod. To support us, Join our Patreon for extra content or donate to Ko-fi.com/romanistan, and please rate, review, and subscribe. It helps us so much. 

Follow Jez on Instagram @jezmina.vonthiele & Paulina @romaniholistic

You can get our book Secrets of Romani Fortune Telling, online or wherever books are sold. Visit romanistanpodcast.com for events, educational resources, merch, and more. Please support our book tour fundraiser if you can. You can give our book a positive review too!

Email us at romanistanpodcast@gmail.com for inquiries. 

Romanistan is hosted by Jezmina Von Thiele and Paulina Stevens

Conceived of by Paulina Stevens

Edited by Viktor Pachas

With Music by Viktor Pachas

And Artwork by Elijah Vardo

Support the show

Speaker 1

Happy holidays Romanistan listeners. So we don't have a holiday episode per se, but we do have an episode we recorded a few months ago with Rowena Marin about Google Translate and the Romani language. We're really excited to bring you more information on this hotly debated topic and we're wishing you happy December happy whatever you celebrate. We are so excited to share more content with you in 2025 and wish you all the blessings for the new year and, if you would like to share the love, consider giving us, or any creators you really love, five stars wherever you can review them and write a little something nice.

Speaker 1

It really helps. You could review Secrets of Romani, fortune Telling on Amazon, storygraph, goodreads, wherever you can review books, our podcast pretty much wherever you're listening to podcasts and consider how you might be able to support other creators that you love too, because a five star review and writing a little something nice helps visibility. It helps people find the people you love, but it also helps us make our way onto charts that might be helpful or get other kinds of recognition. You never know the good that a positive rating can do, so we really appreciate your support. This past year, 2024, was filled with experiences that we never really expected to have, and we owe so much of that to you. So thank you so much for being here. We love you.

Speaker 2

Welcome to Romanistan.

Speaker 1

We're your friendly neighborhood gypsies I'm Paulina and I'm Jez, and today we're here with Rowena Marin to talk about the Romani language edition to Google Translate. But first we have a few business announcements to make. Our book, secrets of Romani Fortune Telling, is still available. We are planning a book tour and, as of right now, we're still adding cities and dates, and we would love to have the opportunity to do book events together and also connect with listeners and readers.

Speaker 1

But we need some help. We are paying out of pocket, like we do for this podcast and pretty much everything else we do with Romanistan, so we started a GoFundMe and you can find it in our socials. We're traveling as cheap as possible, but everything is expensive, so we would really appreciate your help making this possible. We also have gifts at different donation levels, like having your name read on the podcast or getting readings with us or signed copies of Secrets of Romani Fortune Telling, so everybody wins. Check out that GoFundMe and our list of cities and dates on our social media and on our website, romanisempodcastcom. On with the show.

Speaker 2

Rowena Marin is a Romani woman. Author of the memoir who Am I in the World and global agency lead at Google in New York. She is the co-founder of the School of Reinvention, a company meant to help Google align with their passions and purpose. Rowena comes from a close community of the silversmith Roma from Romania and managed to succeed in her career through education based in Romania, spain, france, india and China. She writes about her story in order to encourage other young women from close communities to pursue education, grow and define themselves beyond any labels.

Speaker 1

Welcome, rowena, welcome back.

Speaker 3

Thank you, ladies. Such a pleasure to be here with you and I'm very excited about your book. I didn't know you guys were going to come to New York, so plus me into doing something together.

Speaker 1

Oh, yes, I would love that. Yeah, it looks like it'll be Brooklyn, October 6. So we have had you on the show before and in our last interview about your beautiful memoir who Am I in the World, which everyone should read, and for anyone who needs refreshing, go listen to the episode. But just for right now, do you want to remind listeners a little bit about your journey from selling silver in Romania to working at Google in New York City?

Speaker 3

That was a very long journey, a 30 year long journey, but sure I can, uh, I can do a quick uh top line of the journey. And um, yeah, I was born in in Romania, in Bayamare, um, at the border with Hungary, which is a city that's bordering Hungary, super, super small town. Um, in my community of uh, silversmith, my first language was, um, romani. I think we learned romanian just a little bit later in life and therefore the other languages, hungarian as well um, and I was born to a family where, uh, my both my parents were educated to some extent, which was very different for their generation. My father almost graduated university and my mother went to high school. So, but although they were, they didn't really follow the path that they were educated in. They continued the path of our silversmith legacy, which is the men create the jewelry and then women sell them, either in fairs or on the streets. Therefore, they wanted for my siblings and I to attend school and follow through with our education, although it was very difficult because we would travel a lot throughout the country, but somehow they made it work and therefore I finished my studies, thank goodness, because that helped me, in course, many, many ways. Um, I got a scholarship when I was about 25 to do a master's degree in spain. Um, uh, also through a roma organization that is working across europe, um, and offers these type of scholarships. It's called the Roma Educational Fund If anybody wants to check it out, please do and through them I got a scholarship for a master's in Spain.

Speaker 3

Absolutely not an easy decision to make, because I come from a very close community where women not only they don't finish their, their, their education, but they would not leave the community to do something like that. But I did, and then the expectation was to come back to Romania, to my community. I did for a very short time and then I realized that, um, I needed, I wanted something different for my life. I tried to follow the path that was given to me, but I couldn't. So fast forward.

Speaker 3

Almost eight years ago I moved to Ireland with a job offer from Google, which was not at all easy to get. A lot had to happen, but I'm trying to make it short Because I remember somebody asked me you know, in the way you tell the story it seems everything seems like it was very easy. It's easy to get into Google. It's easy? Absolutely it's not, but I worked for it and it was a dream and I like to dream and, yeah, I think Destiny, the universe, helped me to get to these dreams, so I got an offer from them in 2017. I moved to Ireland. I was there for almost five years and I moved here to New York, where I am now, because in those five years working at Google in Dublin that's where our headquarters is I got married to Roger Osorio, who is my husband, and we moved here after I got the work permit. We moved here, which was three years ago, and that's how I got to Google US.

Speaker 2

Yay, how did Google Translate come to include the Romani language and why? Why did you pitch this, or what was your hope and vision?

Speaker 3

Yeah, it was a journey like with everything else. It wasn't, you know, something that just happened all of a sudden. Um, definitely a process, but it had a start. So I'll start there. In romania, um, the agency called mccann, that's the local agency in Romania. They started this very interesting campaign in our country which was called the Internet Learns Romania in 2002. And the whole point of their campaign which, by the way, was created with many NGOs in Romania was to put on Google as many translated texts from different from Romanian to Romani, with the idea of at some point, with the idea of at some point, the system to learn the Romani language. So when I saw that video and actually, if you guys want to find it online, it's called the Internet Learns Romani. That's the name of the campaign and there's still a video that talks about that in 2022. And for me, that whole campaign was very thought-provoking and an inspiration, because I asked myself, well, what has to really in reality happen for our language to be, you know, translated?

Speaker 2

um by, uh, the google translate team, because there's a team that decides which languages are added and, you know, make like works on on this right yeah, no, I'm not gonna lie like I totally like fucked around with it for like all day and I was like how the hell did they get all of these words, like whatever you guys did, okay, tell me, I'm so excited, yeah.

Speaker 3

So I really like the idea and, based on that idea, I got back to our team, to the Google Translate team, to understand, like I said, what had to happen. And what had to happen is and what had to happen is that somebody had to propose the language, right To propose the language to the translate team, but also, at the same time, they would need a native speaker to validate. And you were in Google and I was in Google.

Speaker 2

Oh my God, and I speak the language.

Speaker 3

Yeah, which is extremely useful, right. So I learned about all of this because, again, destiny, the person that I approached which could have been anybody in that team, but I approached Isaac Caswell, who is the main lead of this translation.

Speaker 3

So he's the software engineer behind the entire launch. So across Google, if you search for the, you know the the newest languages that were added to Google Translate, there have been 110 new languages. So not just the Romani language was added this year, but 110 new languages. So Isaac is the person that was behind this whole project and he agreed to work together and start oh, sorry, sorry.

Preserving Romani Language in Technology

Speaker 1

Wow, that's quite an adventure. Yeah, thank you so much for sharing that and also for sharing refreshing everyone about your journey to Google. I think you know, if you, if listeners, haven't read the book or listen to the interview where we talk with you more about it, definitely do, because there's so much more to talk about Um. So some of the concerns that um that some listeners had because we asked social media what they wanted to know, um, they were very worried that the language is being stolen by AI. They want to know more about how the processes. You had mentioned that, as um a native speaker, you know you were working on it, but they want to know more about who's working on it. Are people going to be continued to be compensated versus the computer just learning how to speak Romani without a lot of human intervention?

Speaker 3

Got it Well. I can't honestly speak to all those aspects. What I can speak to is what actually happened and my involvement in this process right. So, like I said, I reached out to Isaac and I offered to validate the texts as a native speaker, but to answer your question about how the process works. So I'll walk you through step by step how the Google Translate team works on adding a new language and how the process is for translation per se. So the way the process looks like is that the Google Translate uses a method called zero shot translation. You can search for it online zero shot translation.

Speaker 3

It offers this looks like is that the Google Translate uses a method called zero-shot translation. You can search for it online zero-shot translation. It offers a lot of details, but this is a method that they use to build neural networks that can translate between different languages. So for any new language, the team first needs to find text like articles, books, websites in that language, and that will be used as building blocks to train our translation models in a process that we call LANG ID. It can be trickier than it sounds because it depends on how common the language is and what the characters are that it's written in. But once we have enough data for that language, then the team adds the 4,000 most common English words and phrases translated into the language by human translators. With that data, the team then works to train the computational model that handles the translation.

Speaker 3

So, while translation for around half of the new languages the 110 that I discussed works best with the traditional unsupervised machine translation models, for the other half, advances in AI, like our palm to a large language model, have been crucial. So adding a palm to this large language model that I'm talking about was kind of what made it all happen, what made launching 110 languages, you know, be possible. And then, once refined, the translation model is ready for the final phase, which is testing with real life speakers, and the real life speakers could be just external experts in these languages and sometimes, and even more helpful, it could be just native speakers, like I was in this case. So that's the overall process, that was my involvement, and, again, it's all available online. There's actually a blog announcement where we talk about the process for all the languages.

Speaker 1

Hi, it's Jasmina with a commercial. Happy New Year. Through the month of January, I am offering 2025 guidance and prediction readings that you don't have to book an appointment for. You can just send me $40 plus anything else that you want to know about in the new year if there's something specific you want guidance on, or you can just leave it open and see what comes up for you and I will send you in return a 15-minute audio file of your reading or, if you prefer, a written pdf of your reading or a handwritten letter mailed to you. So three options. Each is $40 and I'm offering this only through January. It's a lovely way to get a little guidance and a little bright light for the year ahead. Paulina and I are also always taking new clients. We offer online and in-person readings, so check out Paulina's website, romaniholisticcom, to book with her, or my website, jesminavontulacom, to book with me, or my website, jasminevantilacom to book with me.

Speaker 2

Happy new year which variations of the romani language are included, and how was that decision made? So will other regional variations be added like later on? Is it always developing too?

Speaker 3

it is always developing um, but, to be honest, I don't know how the future looks like specifically for the romanian language. But, um, as things will progress, um, I will definitely be very close to the um, to the evolution. It's just that right now there is no roadmap for that. But what I can tell you is that the um, as we know, the roman language has so many dialects, but the models um that were used to produce the text uh used uh the southern blacks romani, which is actually a commonly used dialect online, and then there were elements of Northern Blacks and Balkan Romani.

Speaker 1

These were the three dialects that were used to train the models and, like I said, I don't know the roadmap for the next steps, but I'll be close to that and I'll keep people informed yeah, you had mentioned that there are so many different linguistic variations of the Romani language, and so another listener question people have been concerned that Google Translate, only representing the more popular Romani language variations, will erase less commonly spoken variations of the Romani language over time. Do you have any thoughts or opinions on that concern?

Speaker 3

Well, we can predict the future, but I'm just not sure, personally, how that could be the case, because google translate, for now on, right now, only represents online. Thus, for instance, let's take spanish. Let's take spanish. Spanish language is spoken in so many countries. Therefore, there are so many dialects. The Spanish spoken in Cuba is different from the Spanish spoken in Chile, and so on and so forth. There are many variations in the words, the dialects and so on, but Google Translate only has two variations of Spanish, which is the Spanish spoken in Spain, and then they call it the Latin version, latin American meaning. So did that erase or change or modify the Spanish that's spoken in Cuba? No, I don't think so.

Speaker 2

No, I think I just want our listeners to know I don't think so. No, I think I just want our listeners to know I am very for personally popularizing our language, um, but I do understand how people could not want that to happen.

Speaker 1

So yeah, I'm sort of of the opinion as a non-native speaker. I didn't grow up speaking the Romani language. I'm mixed. I sort of like don't even really want to share my opinion because I'm just like it's a language I'm coming to as an adult that I'm reclaiming. I'm way more interested in what Paulina and Rowena think about it.

Speaker 2

No, because, listen to me, everybody matters. Just because you're mixed, it just means, like your opinion still matters because, um, it's, it's everybody combined. Like the romani language to be out there isn't for people that have grown up with it and that know it natively. You know the romani language is for people, I think for everybody, like everybody can benefit, especially you. So your opinion matters a lot. J I'm just saying you make sense.

Speaker 3

I agree with that, all right no-transcript.

Speaker 2

Use Google Translate to race fake, a different type of infiltration. Also, some Roma have comfortably written into Romani language on social media to express political opinions or arrange protests and other activist activities, without their words being privy to ai and other tech. At the same time, we recognize that efforts to translate romani literature and other essential documents and preserve the language and its many variants for future generations are critically important. As roma progress, we become more visible. Um, what would you like to share with listeners who are afraid of the repercussions of Google Translate? Offering you know Romanes?

Speaker 3

So I acknowledge that some people might be fearful, and I can also understand. At the same time, you asked me what would I say to all of that? I just don't know where the or the extent to which the Romani language is secret, given that in Romania we have Romanis taught in schools from second grade to university, From second grade to university. So what I would tell people is that at least in Romania we have teachers that dedicate their entire life to preserve the Romanian language and to pass it on to the next generations, and that was my inspiration.

Speaker 1

They are my inspiration. Yeah, I really love that perspective because I think, um, there are different levels of accessibility to the language depending on where you are, and I think, yeah, maybe some of the readers who are feeling or listeners who are feeling really afraid of that might not have grown up with the language being particularly accessible, so maybe it feels very secret. I also think it's interesting the idea of it being the last defense against racism, because Romani people experience racism every single day, and so I didn't really quite understand that fear myself, but I also understand why people feel very protective of the language and feel like it's something that, um, they might want to keep private, um, but it also seems like this is part of progress as being visible, um, so, yeah, I don't, I don't think we get to have both you said it so well.

Speaker 3

Just, I don't think we get to uh, have both, and I read a lot of comments that I think stem from fear and, like I said, I acknowledge I come from a close community. I experienced racism and anything else that we all experience and, like you very well pointed, I don't really understand how our language was a defense, especially. That is not a secret at all. I mean, there are so many texts online that are translated already and, like I said, it is being taught not only to Romani people in Romania but to anybody that is interested.

Empowering Romani Language and Technology

Speaker 1

And our last question. Thank you so much for your thoughts on all of this. I know it's been quite a process and you've been listening to a lot of people's thoughts. What is your hope for the future for the Romani language and technology and is there a way that you think this could create more opportunities for Roma?

Speaker 3

Yeah, so technology can be used in both ways. It can be used for creating opportunity and it also can be used for evil, right, but what I would say is that I think that the roma people, generally speaking, could benefit from technology, and more in a more meaningful way than what is happening today. Google translate exists not just for you know, the ones that are using it for their money people but in general, because people want to interact with other people in their languages, because they want to communicate to others in a deeper way, because they want to understand their spouses, their family members when they speak in their native language. I process, my sister-in-laws would love it if they could understand.

Speaker 3

You know to some extent, what we talk about when we have family reunions, because they're not Roma. So I feel like technology is a tool that can help, can elevate, can grow and bring people together and connect us in ways that we could not even think possible just a few decades ago. And but can it be used for evil that we could not even think possible just a few decades ago? But can it be used for evil? We know that is a possibility in general with technology Not just this one, but any type of technology could be also used for evil. But as humanity progresses, the hope is that we will use technology for the greater good, and that is my hope for this case as well.

Speaker 2

I agree and thank you so much, romina, even from everyone who wanted this, and definitely for me personally, like thank you for putting this in motion and making this happen. It's a big deal, it's a game changer for history, and I'm even happy to be a part of it thanks for it means.

Speaker 3

It means a lot to me. Um and I think I say this in another interview interview, but I do want to stay here as well um, after it was available online. So many of my cousins call me and we were talking about this. My cousins grew up with me in the same post communist Romania where the Roma people were yeah, if we could have disappeared from the face of the earth. I think that's what the Romanians wanted in that moment in time and when they saw that this was available, the reason why we were all so happy was just to see the progress, from where our country was and the way we were seen and what was happening 20 years ago, 30 years ago to now, where we have our language translated. We see progress in this and I hope other people feel the same way.

Speaker 1

I do. Yeah, thank you so much for talking with us about this. I know personally, I get very, very excited when romani literature is translated into other languages. I think that's one of the most exciting things. I always wanted to be a writer and when I was a kid I had no idea that roma wrote like. I just didn't know. And when, I yeah, paulina's hands are both up in the air Like yeah.

Speaker 2

I have no idea. Even up until we started doing this podcast, I didn't know.

Speaker 1

Yeah, and it was like my whole. It was my whole ambition, and I just had no idea that Roma literature existed. And this is like shocking, but that's just how it was. I didn't know until I started grad school and I was like I wonder if there's an anthology out there, and I found the anthology Roads of the Roma. I think it was published in the 80s and there are more anthologies now, but back then there were very few.

Speaker 1

This was like 2011, I think, and I was like, oh my gosh, and I discovered Romani literature and so much of it had been translated from Romani, but also from German, spanish, romanian, like all these different languages, and so the idea of art and other documents and all these important texts being more available to people is exciting to me.

Speaker 1

I would love to see Romani literature blow up. We have so many cool writers and it's hard to find translators. Sometimes that work isn't always really funded. So I was excited about this idea that people were interested in translating Romani language and for literature you'd still need to hire a translator. You'd still need someone to be a writer and an artist to make it sound the way it's supposed to, and so I don't really see this necessarily as like taking away opportunities. I think it's more showing demand, and these are the kinds of things that writers, when they're trying to get a book deal or a translation deal, they can point to and be like look, there's, there's demand, it's on Google Translate and people get contracts this way. So I'm just, I really would love it if we could roll with it and kind of like see the positive in it and make more happen for us, and kind of like see the positive in it and make more happen for us.

Speaker 3

I truly, truly hope that will be the case and that this is building opportunity and nothing else, because, to your point, at least in the country where I come from, there have been no companies that that deal with translation that suffered because google translate exists. You know no type of translation or business that's related to translation has suffered. It just made their, their work, maybe a little bit easier. And also, to your point, it was a like a point to make a business case towards even having such an opportunity. So, yeah, I do hope people see it that way and that they embrace it, and I hope it's useful for people's lives. That was definitely our intention.

Speaker 1

Yeah, and you did so much really crucial work, so we're just um hope. It doesn't sound weird to say we're really proud of you. You're very impressive.

Speaker 3

That means a lot to me.

Speaker 2

It means a lot to me yeah, I look up to you personally also. Um, I do have one question, oh. Um, when are we gonna get the pronunciation feature? Or I volunteer?

Speaker 3

oh, um, it will come. I think it will come. I like I really don't know the roadmap, but what I can tell you is this, and I've been saying it in many interviews it is for people that use google translate please use the um feedback feature oh okay, I'm gonna say that please do because we, the team, really looks at that.

Speaker 3

So when you translate something, um, you'll have a button that's in the results page. So on the right you can barely see it. It's a really, really tiny link in gray that says send feedback. So if something doesn't sound right, or even if you want to say that, look, I want the feature for pronunciation for this word or in general, please use it and send that feedback.

Speaker 1

Thank you for that tip. That's a great idea. Yay, well, thank you so much for spending time with us. It's always so nice to see you virtually and in person, so hopefully we get to see you in person in New York.

Speaker 3

October 6th. I'm going to put it on my calendar right now.

Speaker 1

Um brooklyn right yeah, at grand street healing in brooklyn and for listeners um I don't know.

Speaker 2

I don't know if I'll be there yeah, I.

Speaker 1

I will definitely be there. Paulina might have um responsibilities elsewhere I might, I might just show up depending on how the gofundme goes, paulina may be there. So if you want Paulina to come to New York, you can go to our GoFundMe and our social media. Where is it again? In Brooklyn, where Grand Street Healing. It should be one to three in Brooklyn and we will have a list of their places that we will be.

Speaker 2

Oh, the baby, Baby. I love that so much.

Speaker 1

Yeah, so you know, normally I think we would read out like a whole list of places that we're going to be, but I'm actually going to put the link in the show notes and on our social media, just because it's not finalized yet. And you know what we're doing all of this on our social media just because it's not finalized yet. And, um, you know what we're doing all of this on our own. And if we had like a um and a manager or something, this would all be set in stone. But that's not.

Speaker 1

That's not our lives yet um, okay, great well, have a good day. Thank you, listeners, for tuning in and thank you ladies for the opportunity.

Speaker 3

Okay, great Well have a good day. Thank you, listeners, for tuning in. Thank you, ladies, for the opportunity.

Speaker 2

Thank you for listening to Romanistan Podcast.

Speaker 1

You can find us on Instagram, TikTok and Facebook at Romanistan Podcast and on Twitter at RomanistanPod, To support us, join our Patreon for extra content or just donate to our Ko-Fi fundraiser, ko-ficom backslash romanistan and please rate, review and subscribe.

Speaker 2

It helps people find our show. It helps us so much. You can follow jez on instagram at jasminavantila and paulina at romani holistic. You can get our book Secrets of Romani Fortune Telling online or wherever books are sold. Visit romanistanpodcastcom for events, educational resources and more. Email us at romanistanpodcast at gmailcom for inquiries.

Speaker 1

Romanistan is hosted by Jasmina Von Tila and Paulina Stevens, conceived of by Paulina Stevens, edited by Victor Pachas, with music by Victor Pachas and artwork by Elijah Vardo.