Is human language at an inflection point?
Linguist Adam Aleksic on the impact of algorithms on how we speak.
Embedded is your essential guide to what’s good on the internet, written by Kate Lindsay and edited by Nick Catucci.
If algorithms mark a language inflection point, is AI going to be the next one? —Kate
New on ICYMI: Last week’s piece about logging in plus this week’s dive into the Gen Z stare inspired us to go long on the pod!
Head over here to subscribe to ICYMI wherever you listen to podcasts 🫶
Having helped popularize two viral phrases myself—“millennial pause” and “rawdogging” (“Gen Z stare” up next?)—I’ve long had a particular interest in the neverending group project that is “the words we use to talk to each other.” To me, the most interesting stories about the internet are about how it changes us offline, and there’s probably no aspect of humanity that’s been more changed by technology than language.
And yet,
is one of only a small number of linguists who have turned their focus to the internet. The 24-year-old studied linguistics at Harvard and began making TikToks about it on the side. Now, with his debut book, Algospeak: How Social Media Is Transforming the Future of Language, he’s providing a crucial record of a new era of language that’s changing faster than we can keep up. In this interview, Adam and I talk about how algorithms are changing how we talk IRL, why new words are getting funnier, and the importance of studying all of them, no matter how ephemeral.
You first decided to use your linguistics degree to become a content creator. How did you start making videos?
So, an embarrassing fact about myself is that in high school I was a Reddit influencer.
Perfect.
It was a fun game to see what's going to go viral and what's gonna capture people's social attention. So I knew it was possible. I knew I could combine that with my other interest in linguistics. It's still the underlying mechanisms of what captures our attention which is a whole other rabbit hole that I talk about in my book. And I started being diligent about it, seeing what works, and you can see your video analytics, you can see where people fell off watching, and what do I need to change? And then through trial and error, you get to make better videos over time.
I've interviewed a ton of influencers for this book, and about half it seems are extremely analytical as well and are looking at these analytics. And the other half are just like, vibing it out. You gotta remember there's huge survivorship bias in what goes viral. Some people just happen to, but there's also a mimicry on some subconscious level, like you just copy strategies that you see working for other people. So even the people who aren't being intentional with it are still subconsciously replicating these strategies.
In terms of the internet's effect on words, when did that combine with your linguistics? Like how did you get interested in specifically the internet part?
Well, there's this great book everybody needs to start with, Because Internet by Gretchen McCulloch. And David Crystal has also done some cool stuff on this. He's a British linguist. But there's a burgeoning field of studying how language is used on the internet, and I would say it’s one of the most understudied fields. It is a huge expanse of information, and yet some linguists are hesitant to look into it because it's not regarded as seriously, or it's too ephemeral. If I start writing a paper on the word “yeet” and then it dies out, what use was the paper? But I still think it is worth writing that paper because it's something to focus on how quickly words are turning over.
Like you said, it's like a burgeoning field. What are you hoping your book contributes to it?
I strongly believe there's a few inflection points that changed the direction of language throughout history. When we start writing stuff down onto stone and clay, when we start moving to paper, these affect how we tell stories and how we relate to each other. The internet was another inflection. And Gretchen McCullough covered that incredibly. I think algorithms and algorithmic social media is a new inflection point. It's the driving factor for language change, the fact that influencers are molding their speech around [it]. And to a certain degree, we've always molded our speech around attention, but algorithms are compounding natural human language change. And I think we're having more slang appear than ever before. We're having our language evolve uniquely through this algorithmic lens, and all words are affected by it.
We're talking to a different audience because of algorithms, and so that’s gonna change our speech.
I've thought about that audience thing a lot. I don't really make my videos with an audience in mind. I kind of make it for myself 'cause I like doing the video on some intrinsic level and then I make it for the algorithm and just let it go. And I think it's good if you don't pay attention after that, but that means I'm not even making my video for any specific person. It lands on your For You page, you think it's for you, and then you personally resonate with this video on a more subconscious level. But also it can mean that there's context collapse. So if I'm a left wing creator stitching the video of a right wing creator, I expect it to be served as kind of a rebuttal to the original right wing creator's audience. But the algorithm's gonna reinterpret that. The algorithm's gonna say, “Oh, you're saying something progressive in this video, I'm gonna send it to a completely different audience.” Neither reach is my intended audience, nor are they aware that they're not the audience. Similarly, all the time, African American creators are speaking to their ingroup and then they use a slang word. Now, somebody hears that slang word, they think it's for them, and they replicate the word. And now, now we have way more African American English spreading a lot faster than ever before. And I think it's partially 'cause of that context collapse phenomenon.
We subconsciously replicate formats in language on TikTok and there are moments that I think it doesn't quite work. There was one where this woman who I think maybe makes content about having some type of disorder that causes seizures, but basically the video is her being like, “come with me as I have a seizure in a five star restaurant.” And all the stitches are like, “Oh, no thanks. I don't want to do that.”
So there's a few things going on with those templates. One of them is a framing effect. It puts the viewer into a receptive headspace to watch a certain kind of video right at the start. It's replicating this understood format and there's a lot of that. Beyond the “come with me” and “get ready with me” and whatever, there's certain sentence structures that are now affecting how we talk offline. The way you blank, not you blank. Like these are all from TikToks, originally African American English, but because they're phrasal templates that are catchy or intuitive or work online, they now spread offline.
I confused my mom the other day 'cause she said something and I responded like, oh, not this thing. Like, “not blank.” And then she was like, “Wait, what do you mean ‘not blank’?” And I'm like, “Wait, no, sorry.”
You have to be millennial or younger to understand that. It's giving language change.
You mentioned also that people are maybe reluctant to study this because we have a lot more ephemeral words. Have words been ephemeral before or is this particularly heightened with the internet?
Well, memes and fads have always been around. You go back to ancient Rome, they were carving this mysterious Sator square in the walls. This Jesus fish, they had symbols that would show up in graffiti and they would replicate them. So we know that from graffiti, but definitely just person to person, it makes sense that we are humans, we like trends. I think the internet takes these natural trends which might span the course of decades or centuries in the past, and it compresses them into something intense and commodified and oversaturates us with those trends such that they die out a lot faster.
What makes people stop saying a word?
I talk about that in one of my book chapters. It's an important question to think about right now. There's a few things: How much a word is sticking out, it's called obtrusivity. If a word’s in your face. I don't know if you've seen Mean Girls, it’s why Gretchen couldn't make fetch happen. It was sticking out too much. Versus if a word's flying under the radar, you're less likely to notice it. So here's an example. In 2013, right when the word “yeet” started going, so did the word “selfie,” right? We never thought about selfie as slang, it didn't stick out. And also it more intuitively maybe fit the context, although I would argue that yeet is pretty intuitive. There's a lot of things, it's hard to make a prediction. You gotta listen to what the middle schoolers are saying, I think.
Is there a trend to the type of words we've been creating? Like what does it mean for a word to be defined by an algorithm?
There's no singular thing. There's things we can notice that are general patterns. I think the words we're using just sound a little bit funnier, because the words that are spreading, perhaps the slang words, are memes as well. And the algorithm takes memes, blows them up as viral trends, and then either they find a way to stick around or not. You can't tell whether the memes are actually getting funnier than they were. But I think there's more of a joking mentality online than we've had with a lot of language change in the past.
I feel like the memes are funnier because they require so many layers of context. And if you manage to check off every layer, it's almost like what you're laughing at is how crazy it is that you understand it at all.
Right. There's also maybe a sense of in-group satisfaction. But let's look at a word like “unalive,” which is how I open the book, 'cause I start with internet censorship. It's just a kind of a funnier way to say “kill.” But now we have middle schoolers actually saying unalive. And then we have the obvious brain rot ones, “rizz,” “skibidi,” they sound silly. But then we have more like the under the radar ones, like two words that were widely popularized in 2021: “function,” for party. It's always been a formal event, but now it's an alcoholic rager. And then “side eye,” like looking to the side, those words were popularized by memes. And I think they're a little bit more under the radar and more likely to perhaps survive than rizz and skibidi, which might be on their way out now.
It's interesting how much humor is a part of language, especially when you think about how comedy is considered frivolous. I would say that so much of our language right now is determined by what we find funny.
I think it's 'cause you can't separate language change from algorithmic memes. Like the algorithm is designed to just push these memes because that's an effective way of keeping your attention so they can sell you things in the end. Everything about social media is to hold your attention. There's a few layers going on. There's the meta app layer where the app itself is designed as an intentional capture, and then influencers are incentivized to also find ways to capture your attention. Now we perpetuate words and trends that we know are trending, and we feed into the cycle of those trends kind of paradoxically.
You touched on this before, but what are some drawbacks to this rise in internet-specific language?
I think we lose a lot of context quickly, which can be dangerous. Another huge cultural exporter of new slang is the incel community. So like “sigma,” “Chad,” “alpha,” “mewing,” “mogging,” “pilled,” “maxxing”—these all come from not just the manosphere, but this very, very misogynistic, concerning community. But they're extremely good at weaponizing their ideas through memes and have been able to spread this across different platforms to the point where they became viral ideas. And if you look at TikTok, like the beauty side of TikTok, they'll talk about incel talking points. They'll talk about “canthal tilts” and “interocular distance” and “hunter eyes.” Those concepts all filter down from this weird phenological space on 4Chan in the early 2000s.
There's also a lot going around about the "TikTok accent."
That's chapter four of my book.
What are your thoughts on that? Is that as real as people have been saying?
I think there are TikTok accents, but I don't think there's just one. The most stereotyped one is the beauty influencer lifestyle vlogger, “Hey guys, welcome to my Zoom call,” the uptalk, you'll have the lengthened vowels, the vocal fry derived from the valley girl accent. And it just comes from the YouTube voice of beauty vloggers on YouTube. And then those were just copying the Kardashians or Paris Hilton 'cause it sounded cool. But they are also fundamentally attentional mechanisms because the uptalk keeps you hanging on each word because it makes you feel like something's coming next, and the lengthened vowels are [because] any dead space is very, very bad in videos. Especially when an influencer is working in an extemporaneous or improvisational capacity, they have to come up with ways to lengthen their words so you keep your attention on them. 'Cause if they stop talking, it's over. You're gonna scroll.
I use a very different one. I do talk quickly, but I'll talk very quickly with certain intonations. I know stressing certain words will draw you back into the video. And then if you look at Mr. Beast, right? He doesn't talk the same if you look at an interview of him versus how he talks in his videos. He completely exaggerates it. He is screaming at the camera. It's an intentional mechanism. It's a train wreck. You gotta keep watching it. You can't scroll away from this intense affect that he puts on. And there's a few others. There's the ASMR voice.
There's a question that maybe we're touching on here: Are we losing offline accents? And this has been a trend that's been happening since globalization before even the internet. We've already been losing regional dialects, especially all the local dialects in England. There's a new study that there's a 60% reduction in the Texan accent, and a lot of America is trending toward this "neutral" midwestern dialect. I think the algorithm might be accelerating this, but at the same time, there's a lot of cool language preservation efforts on the internet. And I do think maybe it's true that the offline accents are being replaced with online accents. We're simply spending more time in this online space and we're supplanting that with this new breadth of linguistic diversity. And I do keep coming back to it, like there is a homogenizing effect that all these are for attention.
Adam’s book Algospeak: How Social Media Is Transforming the Future of Language is out now.
P.S. If you enjoyed this dive into the internet and linguistics, we have an ICYMI episode all about it from earlier this year!