225. Speaking Fluent Internet: How Algorithms Are Changing the Way We Speak

[00:00:00] Matt Abrahams: Words and language are conduits for culture. My name's Matt Abrahams, and I teach Strategic Communication at Stanford Graduate School of Business. Welcome to Think Fast Talk Smart, the podcast. Today, I'm delighted to speak with Adam Aleksic. Adam is a linguist and content creator posting educational videos under the name the Etymology Nerd, and he has over 2 million followers. He's lectured on language and social media at Stanford and other top universities. His latest book is called Algospeak: How Social Media is Transforming the Future of Language. Well, welcome, Adam. I look forward to our conversation. My younger son is a follower of yours, and he and I are both really excited to hear this conversation. Thanks for being here. 

[00:00:46] Adam Aleksic: Hi. Excited to be here. 

[00:00:47] Matt Abrahams: Thanks. Shall we get started?

[00:00:49] Adam Aleksic: Go ahead.

[00:00:50] Matt Abrahams: As the son of both a lawyer and an elementary school teacher, language and words have always been really important and interesting in my life. I'm curious, where did your interest and passion come from and why should any of us care about the words we use in our grammar?

[00:01:04] Adam Aleksic: If I'm being honest, I got into it for the fun facts. I read this book in 10th grade, The Etymologicon, great book, connecting how different words are all related to each other in different ways. Fascinating. And then the more I looked into it, the more I started doing my own research and eventually studied linguistics in college. I began to realize that language is the way humans identify the world and relate that to each other. It's so much more than just fun facts. But those are still good, you know, in and of themselves. It's how we communicate, who we are as people. And you can see this as a, language as a proxy for culture.

[00:01:34] Matt Abrahams: Absolutely. And if you have a word to identify it, then it exists and you can share that meaning, right? 

[00:01:39] Adam Aleksic: Exactly. But once it's identified, it also changes reality a little bit. The fact that a label is out there. You identify either with that label or against that label. If that label wasn't there, now it's not something affecting your identity. This is something that's very interesting to me 'cause the book that I wrote gets into algorithms and how algorithms create new words. And you have new labels out there because algorithms push these, like cottagecore or something. Now that cottagecore is out there, I'm either cottagecore, I'm not cottagecore, and that affects my identity subtly.

[00:02:04] Matt Abrahams: I have no idea what cottagecore is. Will you share with me? 

[00:02:06] Adam Aleksic: It's a fashion aesthetic of like, it's sort of a bucolic lifestyle of prairies and whatever, and you're dressing toward that aesthetic, but it's packaged as an entire lifestyle on social media. 

[00:02:14] Matt Abrahams: Interesting. So when Spotify identifies a particular type of music or labels a type of music, all of a sudden it now exists and you're either it or not it.

[00:02:23] Adam Aleksic: Exactly. For this book, I actually interviewed Spotify's Chief Data Alchemist, Glenn McDonald and they have over eight thousand micro genres on Spotify. You know, in the past we'd have something like R and B, hip hop, pop. Now it's also escape room, preverb. These are like things they made up because they're just putting new labels on things, and now artists try to conform their musical identity for these labels that are out there. Hyper pop is the term that wasn't really on the scene until Spotify created a hyper pop genre in 2018. And now people are trying to make hyper pop, they're arguing over what is and what isn't hyper pop. But they wouldn't have been doing that otherwise. They would've maybe just been making music in this space and more fluid perhaps, but maybe the label constraints identity. At the same time, maybe the label gives people a way to pinpoint exactly what this thing is. So what you were saying, words have tremendous power. 

[00:03:09] Matt Abrahams: That's fascinating. The, the words we use not only create and identify what exists in the world, but how we see ourselves as being part of or not part of. So let me ask you talk a lot about algospeak. What is algospeak and how is it impacting our communication? 

[00:03:26] Adam Aleksic: Traditionally, algospeak has been the label for speech used to circumvent algorithmic censorship online. The classic example is the word un-alive. You can't say kill on TikTok or you can say it, but your video might be suppressed. So many creators choose instead to use words like un-alive, and now we have kids in middle schools writing essays about Hamlet contemplating un-aliving himself simply because they see it on TikTok. And so now this is also changing our reality as well. That's traditionally been what's referred to as algospeak, and that's what got me interested in this because I'm a creator and I'm a linguist, and I started studying my own language because you can't not if you're those things. And I started realizing, wow, my speech is being hampered and by the algorithm I have to reroute around it. At the same time, it's so much more than censorship, like un-alive. It's memes, it's trends, it's which groups are created, it's where words come from in the first place, and how quickly those words spread. 

[00:04:15] Matt Abrahams: Wow. So the algorithms are replacing the traditional way that new words came into being, subcultures, using concepts, and ideas. Given that there's this new language, if you will, or at least new words in a language, how do we best become fluent? I'm thinking of the modern day workplace where you have four generations of folks, and some folks are religiously on TikTok and Instagram and part of that algorithm culture. And then you've got folks who are older who don't even know what those things are. How can we expect to communicate when we're not even speaking the same language in some cases? 

[00:04:50] Adam Aleksic: Right. Each social setting comes with a unique set of linguistic expectations. In the workplace, like you described, there's still like a general corporate kind of language or an attitude about what words are acceptable and not acceptable. Same with un-alive, right? It seems unacceptable for a kid to write that in an essay. But it's acceptable to say on TikTok because that actually is the expectation, that's the sociolect, the social kind of dialect of the internet, that people have this certain way of speaking that finds a community. 

[00:05:15] Matt Abrahams: So if we are to be a leader in an organization, a manager of a group, and I want everybody to feel included in that group, do I have to have a glossary of terms that we agree to, to use? If I've got a colleague that refers to something I don't understand, am I being left out? Am I being ostracized? What recommendations do you have or have you seen so that we can all communicate the same way? 

[00:05:39] Adam Aleksic: Well, I think a lot of these expectations are implicit, and it's strange to make a list of words that are acceptable and not. There are key words that are being flagged by the algorithm, or some words the algorithm knows are trending, and then the algorithm pushes them and then creators use these words, and then the word gets perpetuated further into virality, and now it's more of a word. That to me is also algospeak. I'm a strong believer that the medium is the message. Every single medium uniquely affects how we communicate and we adapt our speech to these medium.

[00:06:05] The algorithm is one such medium, and we have a unique way of speaking for these because algorithms are an infrastructure underlying how videos get distributed. You have to appeal to this algorithm, not only in which words you're using to avoid censorship, but also in which words go more viral. But also, generally, if this is a social expectation that we should be speaking this way, you have to accommodate for that as well. In a office setting, the medium is different. You should not be using the word un-alive in the office, and that's just not the expectation. 

[00:06:32] Matt Abrahams: So each context brings its own expectations, which brings along with it the certain rules that you have. I find it really interesting. One of the most ubiquitous bits of advice that comes from this podcast is you need to know your audience and you need to adjust and adapt. And I'm hearing that you're adding a new audience that some of us need to consider, which is the algorithm and the tools that we use. And that's really interesting because if I want to be understood, I need to at least speak the same language or understand and appreciate the language of the tool that I'm using.

[00:07:00] Adam Aleksic: The context is really important. For example, right now, I know I'm talking to you and I'm accommodating my communication to talk to you. However, we're also being observed by an audience of many other people, and I don't know who these people are, so I'm using a psycholinguistic phenomenon known as audience design, where I have an idea of a perceived imaginary audience and I am accommodating my communication for that as well. This is something we've seen with radio hosts and broadcast presenters. You'll notice that they talk in specific accents, right? Uh, I was on NPR the other day, they were talking about the NPR accent and how they talk like that.

[00:07:33] And we all know, this just in, breaking news, they have a certain way of speaking on TV as well. Influencers as well have a sort of an influencer accent. They're accommodating for their perceived audience. And again, this comes with the specific social setting and the norms and the underlying values of that medium. However, we're also communicating, not only for the actual person in front of us, not only for the people invisibly perceiving us, but also for the algorithm itself. There's also this invisible infrastructure that analyzes every single word that's uploaded to these platforms, and you need to be performing for that as well. 

[00:08:05] Matt Abrahams: I want to come back to influencers and creators in a moment, but two questions pop into mind. AI and the algorithms are only as good as the data that they have come to digest and understand. As language is changing, is it possible that the algorithms don't yet know what some of these words and terms mean and they might actually make mistakes? 

[00:08:26] Adam Aleksic: Not only possible, but that's exactly the reality. The fact is that AI, as sophisticated as it is, has a training set and has reinforcement learning that is a biased version of reality. It contains an idea or representation of language that is a map, but the map can never be the territory itself. So the way human language is actually used in all of its nuances in all of its little forms and the way that it's currently changing, AI can never fully catch up to that. And that's why if you ask ChatGPT to use slang, I don't think it'll ever quite be hip to that. Human language is always one step ahead, which is why we come up with new phrases like un-alive and now that's being censored as well because the algorithm caught onto that. But now creators use phrases like, still unalive, but like spelled with an at sign instead of an A, or an exclamation point instead of an I, one step ahead of the algorithm. If there's one thing I can count on humans, it's that we are ingenious and tenacious with coming up new ways to express ourselves. 

[00:09:14] Matt Abrahams: I wanna dig into your expertise as an influencer yourself, as the Etymology Nerd, but as somebody who studied algospeak. If it is all about getting attention, sustaining attention, leading to virality, so your ideas get heard by more and more people, what types of advice would you give to people who are trying to be really fluent and maximize what algospeak gives for them? Are there some specific tips and tricks that you would recommend? 

[00:09:40] Adam Aleksic: Start with what these platforms want. The platforms give us the infrastructure. And unfortunately, they are trying to grab our attention as much as possible to commodify us as people so they can sell our data and so they can sell us ads. That's what they're doing. They make the structures in place. They have retention as like one of them, that's like how long people stay watching a video. They use engagement metrics and whether they recommend videos, that's like likes, comments, shares, all that normal stuff. So as influencers, you are then incentivized to maximize for these things. It's all downstream of how the medium is baking in certain expectations, certain priorities, and so you need to be very aware of that.

[00:10:15] And it's also, again, be aware of your invisible watching audience as well. I speak differently than the lifestyle influencer because I have a different audience. The lifestyle influencer is probably speaking to young women who expect this sort of valley girl derived accent. I speak very quickly, I mean, I do in real life, but I'll online, I'll stress more words to keep you watching my video to make sure you don't scroll away, and I'll uptalk as well, but it'll be like a different thing than the lifestyle influencer, because I'm also accommodating for both my imaginary audience and the algorithm.

[00:10:41] Matt Abrahams: Really interesting, and you're making me think about what I do linguistically when I speak. I guess I am purposely doing things for that imagined audience. It might not be good for the algorithm, which might not be good for me to get my word out there. Are there certain things people can do that enhance virality beyond just the spoken word? So for example, what we name our posts, which not everybody's an influencer or creator, but I put titles on emails. I label what my meeting is. Are there words, I mean, words, communication as well?

[00:11:11] Adam Aleksic: Written words grab our attention more than other things. I wanna bring in visual semiotics as well.

[00:11:15] Matt Abrahams: Stay, take a step back. Define semiotics for us.

[00:11:17] Adam Aleksic: Oh, it's symbols, things we look at that have meaning. So the fact that we have microphones right now in front of us, and if this is uploaded to social media, there's studies that have been done that show that videos where people have microphones in front of them go more viral than otherwise because they connote authority. This is a semiotic symbol that holds authority within it. And so there's actually an influx right now of real creators using fake microphones because that simply goes more viral. And it literally is a visual indicator that this person has something important to say. And so we are subconsciously primed to perceive that person as more important and as having something worthy when they're talking. And so we watch further. 

[00:11:53] Matt Abrahams: I really find that fascinating. Not everybody, again, is going to be a creator or influencer, but all of us do virtual presentations where we might put a background behind us, and you have to think about what's in that background because that sends a semiotic message, as you're talking about. It's implying things.

[00:12:06] Adam Aleksic: I think culturally we're still grappling with the move to Zoom and the fact that you can just see inside people's bedrooms now. That's crazy. So now your bedroom, this very private sanctum, is now perceived as a professional environment as well. In the same way that we see visual stuff very physically affect how we perceive a message, the same is happening with communication, both written and verbal. 

[00:12:27] Matt Abrahams: I'd like to talk a little bit about the word like. Because it is used in a variety of ways and I'm curious to get your perception of its evolution, and people use like in some ways as a filler, in some ways as to, to replace the word says or said. Help me understand like.

[00:12:45] Adam Aleksic: Yeah, this isn't even algospeak. It goes back to 1980s Valley girl speech and early internet speak, but also like you, haha ha ha. I just used it as a filler word right there. When you said you can use it as a synonym for said that's not quite true. It has its own kind of feeling. Said is more connoting evidentiality, that you literally are repeating something verbatim. Like is embodying an affect of somebody who communicated something but not necessarily saying one hundred percent this is what they said. So if I said, then she was like, that party was low-key great. I'm not necessarily stating that she literally said that. But all these uses kind of evolved out of how young women talked in California in the eighties and nineties. And language often follows the conduits of what is perceived as popular, what is perceived as cool, what is perceived as funny, even. And the Valley girls were popular and they spread language because of that. 

[00:13:38] Matt Abrahams: I grew up in California in the eighties, and I remember Valley Girls speak very much, and it's interesting to see what has lasted from that all these years later. You know, for the past little while I've been studying how we can use ambiguity in our communication strategically. Hedging language, if you will. Think of things like politeness, flirtation, negotiation. All of these are situations where indirect speech, indirect language can actually be very helpful to you. We don't wanna get in trouble. We don't wanna risk our reputation or make a mistake, so we use this ambiguous hedging language. What are your thoughts on ambiguity and how does that play out in this world of algorithms? 

[00:14:18] Adam Aleksic: My first ever linguistics research project that I did as a senior in high school was, I was an intern at a court and I was analyzing how lawyers talk to the judges. And they'll often say things like, your Honor, I would submit this brief, as they're submitting the brief. You don't need to say would, right, theoretically, because you're literally submitting the brief. You could say, your Honor, I'm submitting this brief, but they say would as a hedge. They say it because it shows deference, it shows politeness to the judge. And this is how you'll see a lot of people use this sort of, that would is an example of a modal, auxiliary verb.

[00:14:45] And you can say, I could, I should do this, I shan't do that. All of that is an example of modifying your speech to make it seem more polite. And like is actually, one use of the word like is to hedge. It's showing that you're not asserting yourself too much. On social media, I think it's probably bad to hedge that much because we want to convey authority. That's why the fake microphones actually work, and real microphones, these are real, I'm assuming. But hedging does work depending on different contexts. If you're trying to talk to a judge, maybe it is good to be less abrasive. 

[00:15:17] Matt Abrahams: I see. The hedges that, that bother me the most are, I think and kind of because they diminish your status and power, but at some points, in some ways they can serve a useful function. If I'm the big boss and you're my employee and I genuinely want your input, if I were to say, I kind of think we should do this, what do you think? That lowers my status and the certitude of my assertion, so that you're more likely to say something, versus, we should do this. What do you think? That's a very different sounding phrase, so I find that really interesting. Do you see that play out in your world?

[00:15:49] Adam Aleksic: Yeah. When we're talking about performativity, one thing you're performing for is there's always an imbued power dynamic in any conversation you're having. So if we were like seated on stage in front of an audience, the audience is definitely less powerful or something. Or if you're talking to your boss, that's a perfect example. You have less power than your boss and you need to show deference. And there's languages like Japanese that literally encode this. They have many layers of like different particles you attach to show your relationship to this other person. In English, we do the same thing, but we do that through these sort of hedging particles.

[00:16:16] Matt Abrahams: Interesting. I love learning from you and I, I love the ideas of language. I'd like you to reflect, across all of your work, the book you've written, all of the social posts you do, your own research that you've done, what are two or three key actionable takeaways that our listeners can have from that body of work that will help us in our communication and help us in the language that we use?

[00:16:37] Adam Aleksic: Takeaway number one, each medium is going to uniquely affect how we communicate, and you need to be exactly aware of what that medium is doing if you wanna communicate effectively. Algorithms are a new inflection point for language. I think as a medium, they value attention more than any previous medium we may have had. It's causing language to occur faster, and you need to be tapped into these trends because that's another thing these algorithms are pushing. So be aware of the medium. The second one is that algorithms are changing the way we speak.

[00:17:04] It's a reality that, even if you're not online, you need to be aware of these new slang words, where they're coming from, because these are also conduits of culture, like I said earlier. We know like in the same way that like came from Valley girl English, there's like conduits of popularity, what we perceive as cool or prestigious, and we borrow from that. It's good to be aware of these things so we can communicate more effectively and so that we can be more aware of our own word choice and perhaps the implications of what words we're using.

[00:17:27] Matt Abrahams: I wanna put an exclamation point after this notion of every channel through which you communicate, we have to be mindful and think through the language that we use to communicate for that channel. A lot of us just create one message and then put it out in lots of different channels, and we might be missing an opportunity to optimize. And we might be doing ourselves a disservice because in essence, we're not communicating in the way that we should, and the algorithm is of one big channel we need to be considering. So I'm a grammarian. I think grammar is important. I will defend the Oxford comma. I'm curious what you think about grammar. How does grammar play out? I know in social media people don't use grammar, and in fact, when I text younger people and I put punctuation in, it actually conveys meaning that I'm not intending it to convey. What are your thoughts about grammar?

[00:18:09] Adam Aleksic: Grammar is a set of rules or expectations about how language should be used. This differs on context, and this is where I might push back on you. I agree grammar exists and I agree that when I'm writing my book, I should use the standard grammatical conventions of English. I am an Oxford comma enjoyer in that context. If I'm texting my friend or my roommate or something like that, it doesn't matter. In fact, it would be weird if I used a period when I texted my friend, because here's the thing about texts and how boomers tend to, no offense, hey, they tend to use more periods in their text than younger people. The act of sending a text itself is already a signal that the sentence is completed. So going out of your way to include an extra period is actually saying something extra. You have your sentence and then you have this period at the end of it that you didn't need to send because the message as a whole already communicated that this message ended. So now we as younger people are asking ourselves, why did this person add a period? And we think that's passive aggressive. And so there's this idea of a passive aggressive period. 

[00:19:03] Matt Abrahams: You have convinced me not to put a period. While not a boomer, I do appreciate that. I appreciate that grammar, syntax is important and it, it varies by channel.

[00:19:13] Adam Aleksic: If anything, that's what I mean when I say that every medium really affects the message. It's good to be aware of standard grammar for sure, because there's a lot of context in which that is the way you professionally communicate with other people. But it's also good to be aware of the informal grammars and being aware of all these grammars at once allows us to communicate better across different medium.

[00:19:30] Matt Abrahams: Yeah. But it puts the onus on us to actually take the time to think about that and learn that. Absolutely. So Adam, before we end, I like to ask three questions of everyone. One I create just for you and the other two I've been asking across all the episodes. Are you ready for that?

[00:19:43] Adam Aleksic: Let's do it.

[00:19:43] Matt Abrahams: So I would like, as a very bright young linguist, for you to give a commentary on the name of this podcast because it is grammatically incorrect, Think Fast Talk Smart. It should not be grammatically worded that way. Is this a good thing? Is this a bad thing? What do you think?

[00:20:02] Adam Aleksic: It's funny because, again, grammar is subjective here. We've had an increasing informal dropping of L Y suffixes, the turning of adverbs into adjectives, and vice versa. We've had this mixture. And it's not grammatically incorrect if you're trying to connote this hustle vibe or like, let's get going. Yeah. I think the name of the podcast works for the feel you're trying to create. If you think smarter, you know,

[00:20:29] Matt Abrahams: Smartly or whatever. Yeah.

[00:20:30] Adam Aleksic: It, it doesn't have the same vibe. And different ways of speaking, have different vibes. So I think you're totally fine. And it's grammatically correct for what you're trying to do. 

[00:20:37] Matt Abrahams: Excellent. The high school English teacher that I had, who is upset about the way we named the podcast, I'm gonna send this clip to him so he can understand it. But again, I think your answer shows that the language we use, the words we use, convey more than just meaning. They convey a feeling, a sense, a vibe, as you said. Question number two, who is a communicator that you admire and why? 

[00:21:00] Adam Aleksic: I interview a lot of different educational creators in the book, and I think these people are in a unique position, because you have this desire to communicate something and let people know more about this topic, but at the same time, we're heavily constrained by this medium. You've always needed attention to educate people. A teacher still needs to get her students' attention because that's how you like hold the floor, right? Otherwise, the students will be distracted. However, educational content creators are faced with that unique sort of need to create edutainment, and I'm very impressed with, especially, the early people who had to figure out this space, the Green Brothers, Vsauce. I think they've done incredible work in that regard. 

[00:21:37] Matt Abrahams: This ability to get attention and sustain attention, I define engagement as sustained attention, is critical.

[00:21:43] Adam Aleksic: To me, engagement is a dirty word just because it's like algorithmic social media kind of a word, but you're absolutely right.

[00:21:48] Matt Abrahams: Long before these algorithms existed, I've been teaching this notion of engagement and I'm happy to use a synonym, but there's a need to not only get attention, but to keep that attention and whatever we call that is important. Final question for you. What are the first three ingredients that go into a successful communication recipe?

[00:22:04] Adam Aleksic: I'm a little biased by what, everything I've just said. The first ingredient is just being aware, knowing what the rules are for each social setting. Knowing what the background is of the words you're using. Being aware is just step one always for me. Step two is know the exact setting that you're in. Now you know how these words are used across settings. Now you know what setting you're in, and then communicate conscientiously based on that. Step three, continue observing, become more aware. It's just a cycle. It's, we're performing our own awareness into existence as well. 

[00:22:34] Matt Abrahams: So really it boils down to being open and aware, to seeing what's going on, and then make conscious choices to the language you use. 

[00:22:41] Adam Aleksic: And then you can sort of reflexively respond to what's going on. Continuously, I think reevaluating our state in the world, because here's the thing about language, it's always changing. The mediums are also always changing, and unless, we can become complacent, it's so easy to become complacent. You're like, oh, I figured out how language works. This works great. And then we have a new medium. We have algorithms now. We have new words emerging, and you're no longer aware. And now you're outta date. You're using periods in your text messages.

[00:23:05] Matt Abrahams: I told you I'm gonna stop doing that. This idea of awareness allows you to be agile and allows you to adjust and adapt, and I appreciate that. Adam, this has been fantastic. I knew it was gonna be a really interesting journey. You helped us understand that it is important to think about our audience. The algorithm is one of those audiences. This notion that language is a conduit for culture and meaning is very critical. I appreciate your time and best of luck on your book, Algospeak.

[00:23:32] Adam Aleksic: Thank you. It's so fun talking to you.

[00:23:35] Matt Abrahams: Thank you for listening to another episode of Think Fast Talk Smart, the podcast. To learn more about language and words, tune into episode 91 with Valerie Fridland. This episode was produced by Katherine Reed, Ryan Campos, and me, Matt Abrahams. Our music is from Floyd Wonder. With thanks to Podium Podcast Company. Please find us on YouTube and wherever you get your podcasts. Be sure to subscribe and rate us. Also follow us on LinkedIn and Instagram. And check out fastersmarter.io for deep dive videos, English language learning content, and our newsletters. Please consider our premium offering for extended Deep Thinks episodes, Ask Matt Anythings, and much more at fastersmarter.io/premium.