How algorithms are changing the way we speak

OLIVIER MORIN/AFP via Getty Images

Olivier Morin / AFP via Getty Images

Social media has birthed an entire lexicon replicated by millions online — even if these words don’t actually mean skibidi. On today’s show, we talk to author Adam Aleksic about how TikTok and Instagram's engagement metrics, and viral memes, are rewiring our brains and transforming language at warp speed.

Adam Aleksic’s book is Algospeak: How Social Media is Transforming the Future of Language 

Related episodes: 
What we’re reading on the beach this summer  

For sponsor-free episodes of The Indicator from Planet Money, subscribe to Planet Money+ via Apple Podcasts or at plus.npr.org.  

Fact-checking by Sierra Juarez. Music by Drop Electric. Find us: TikTok, Instagram, Facebook, Newsletter. 

Transcript

ANNOUNCER: NPR.

[COIN SPINNING]

[THEME MUSIC]

ADRIAN MA: One of the beautiful and sometimes baffling things about the English language is that it's always changing. Like, take the word "rizz," which means something like charisma, or the word "skibidi," which kind of means anything or nothing, depending on how you use it. A few years ago, words like these were little known or didn't even exist. But today, they're understood and used by millions of people. Now, don't worry if you don't count yourselves among them, because according to our guest today, the English language is changing at a faster pace than ever before, thanks to social media and algorithms.

[MUSIC PLAYING]

MA: This is The Indicator from Planet Money. I'm Adrian Ma. Today on the show, a conversation with linguist Adam Aleksic. He's author of a recently published book called Algo Speak-- How Social Media is Transforming the Future of Language. This was actually a book I picked for an episode we recently did on econ beach reads. And after the break, we'll talk with Adam about how the business priorities of social media companies increasingly shape the way we speak, both online and offline.

MA: If you are a nerd for words, you might have come across Adam Aleksic's videos on social media.

[AUDIO PLAYBACK]

ADAM ALEKSIC: Fun fact, everybody has a completely emoji style, kind of like an emoji fingerprint, which also means that no two people will use emojis the exact same way.

MA: Using the handle EtymologyNerd, Adam posts videos on TikTok, YouTube, and Instagram, and spends a lot of time breaking down how the internet is changing the way we communicate.

ALEKSIC: In the same way, everybody builds up their own inside jokes.

[END PLAYBACK]

MA: Now, in case you're wondering, you have not accidentally hit double speed on your podcast player. This is how Adam talks in his videos because, he says, it keeps viewers engaged. And when I called up Adam recently--

ALEKSIC: Adrian, hello. Testing, ipsum lorem.

MA: --I started off asking him why he thinks algorithms are shaping the English language like never before. So you say there are these inflection points in history that have had a huge effect on the way people communicate. There's the invention of writing and the printing press and the internet. Now, you say, we are living through another inflection point shaped by algorithms. Can you give us your argument in a nutshell?

ALEKSIC: Every time there's a new medium, we're going to have a new way of communicating through that medium. The underlying process is the same. Humans use language to communicate with one another. However, now algorithms are an underlying infrastructure of the internet and the way we mass-communicate with one another. That means that the way we speak is going to be downstream of platform priorities. Platforms build in attention incentives. We have all these algorithms shaping in groups and echo chambers and where words come from, how words spread, how trends get popularized, how some words are censored, and we have to find ways around that. We've always had these processes in one way or another, but algorithms are affecting it in a new way. So one example I start the book with is the word "unalive." The word "kill" is suppressed on TikTok, so many creators have turned to say "unalive" instead. And now we have kids in middle schools writing essays about Hamlet "unaliving himself," and that's an example of social media algorithmic speak bleeding into the mainstream. And that's traditional algo speak.

MA: Language has the potential to go viral in a way that it didn't before.

ALEKSIC: Look at the word "rizz," which started out in 2022 in the back corners of the internet. Kai Cenat popularized it on Twitch. By the end of 2023, it became the Oxford English Dictionary Word of the Year. I think it might have taken a word like "rizz" maybe a decade to become popular in the past. But because of how algorithms amplify trends, they incentivize creators to use trending metadata because that's a way that creators tap into ongoing fads and stay relevant. Then these words get pushed further. And the word "rizz" entered what I call the engagement treadmill, where some words get pushed more because they are trending.

MA: You reference types of grammatical gatekeeping that happened historically and in modern media. Are these sort of gatekeeping rules just out the window now?

ALEKSIC: Online, there's the experience of context collapse, where somebody thinks that they belong to a community, that they can use this word. And we see the rapid replication of African American English for this reason. There's also no policing function. In the past, there may be people who look at you funny if you use an African American slang word. But now, there's no, like, audience perceiving you in that way. Once you get this video from a different context, you don't even think it is from African American English in the first place. And in the same way, that's just one example of a community that's emanating words. Incels and the manosphere also came up with a lot of words, including the suffix "-pilled" or "maxxing" or the word "sigma." And they're spreading this language, and you perceive it out of context. You don't know that it comes from that group. You replicate it because it sounds funny. At a certain point, we all end up talking like that.

MA: I'm curious, like, being so steeped in this all the time, you must have some favorite algo-derived words.

ALEKSIC: I'm particularly partial to the word "skibidi." It comes from Skibidi Toilet, this YouTube Shorts series where there's a fictional toilet race battling these camera-headed androids, and it kind of speaks to these young children.

[AUDIO PLAYBACK]

SPEAKER: (SINGING) On, on you, on you

[END PLAYBACK]

MA: [LAUGHS] OK, I think I understood about 50% of that. But OK, a YouTube show that pits a toilet against androids.

ALEKSIC: It sort of speaks to young people, but it's seen as funny by older people. And then because we make fun of the younger kids for liking Skibidi Toilet, we push this word further. And then it becomes sort of this, like, hopscotching between irony and what's actually real. And also now influencers are using the word "skibidi" because it's trending, and they know that if they capitalize on this trend, they can go viral.

MA: But it's not all about trends spreading words, right? I mean, sometimes these new words serve a genuinely useful function. You talk about how, for example, the phrase "canceling somebody," it was used in Black culture, but it wasn't widely used until pretty recently. And now, everyone knows it. It kind of fills this semantic gap in our language.

ALEKSIC: An important thing to remember is that words are memes, and they get tied to meme lifespans. If there's no natural reason to keep the meme around, it dies. One usual reason we-- memes do die is because they get used by this outgroup that no longer feels like it's funny or cool. If your grandmother starts saying "skibidi," "skibidi" is going to die out. I'm a strong believer that in about a year, "skibidi" is going to be out.

MA: OK, you heard it here first.

ALEKSIC: [LAUGHS] But cancel is going to stay around because that's a useful term for this new internet concept that we have. Let's say "selfie" emerged around the same time as "yeet" and "on fleek." And yet we don't use those words anymore because they were tied to this obtrusive meme that sticks out. However, selfie, also being a meme, it just has a longer tail because it's now applying to this lexical gap.

MA: Social media and algorithms, you say, also influence the way people talk, like their pacing and their tone. Can you talk about how that happens?

ALEKSIC: Yeah, one chapter of the book focuses on the influencer accent. So I talk quickly in real life clearly, but online--

MA: OK, I was going to ask you about that. Are you in sort of teaching influencer mode, or this is the way that you are when you're hanging out with your friends?

ALEKSIC: There's some intermediate think, right? Every person will accommodate their communication for their expected audience. Right now, I'm trying to cram in a bunch of information so people will buy my book. However, if I'm talking to my friends, I'll use a more laid-back pace because I'm not trying to sell them anything. Online, everybody's trying to sell you attention all the time because the platforms bake in attention incentives so that they're trying to sell you things. They're trying to commodify your data. They're trying to keep eyeballs on the app as long as possible. They make it incentivized to grab your attention, so creators replicate that because we're just trying to earn a living. And then we have all these attention-grabbing tactics replicate. And one of them is the accent. The "I'll talk faster, and I'll stress more words," that's my educational influencer accent. You also have, like, the stereotypical lifestyle influencer accent, which is good at grabbing attention for a different kind of audience, like the "Hey, guys, welcome to NPR." You have rising tones, which make it sound like something's always coming next, so you keep listening. And it also prevents--

MA: [INAUDIBLE] yes.

ALEKSIC: Right. Yeah, it also prevents dead air, which is, like, so bad for the algorithm. So there's new ways of communicating that have emerged, that this is based on stuff in the past that just comes from Valley Girl accents. There's a sort of social prestige that perpetuates with these accents. And the educational influencer accent, there's-- like, there's always, like, floor-holding as a concept in linguistics that people need to make sure people keep listening to them. And we've had attention. If you're a teacher in a classroom, you've got to make sure kids are entertained. In another way, though, algorithms, like, really amplify that natural human behavior.

MA: What's interesting to me about this is that on the one hand, social media is very democratic. Anyone can publish anything they want anytime. And yet, like you say, the way algorithms incentivize certain kinds of posts, the way they drive people to maybe gravitate toward certain ways of talking, it kind of also flattens the way people talk.

ALEKSIC: Absolutely. It does, like you said, elevate more voices in the way that we haven't had before. But at the same time, it prevents those voices from being heard unless they conform to platform incentives.

MA: When you put it that way, it does seem to kind of cast a slightly negative light on the fact that social media and algorithms are shaping the way that we talk. Is that how you think about it?

ALEKSIC: I don't think there's anything ever wrong with language per se. Language is a way that humans relate to one another. I think language is a proxy for culture. And by following the conduits of language change, you can sort of understand more of where we're heading as a society. I do think probably the fact that these platforms are commodifying retention, that seems bad. And you can see that happen through language. But the way we're communicating with one another is just, again, us reflexively adapting to a medium to be heard. And I don't think that's actually bad.

MA: OK. Well, we've been speaking with Adam Aleksic. His new book is called Algo Speak. Am I doing it right, the influencer voice?

ALEKSIC: [LAUGHS] Perfect.

MA: Yeah, Algo Speak-- How Social Media is Transforming the Future of Language. Adam, thanks again for being with us.

ALEKSIC: Thank you for having me.

MA: This episode was produced by Julia Ritchey, with engineering by Kwesi Lee. It was fact-checked by Sierra Juarez. Kate Concannon edits the show. And The Indicator is a production of NPR. Transcript provided by NPR, Copyright NPR.