Apr 18, 2026 · 1h 36m
How algorithms and viral creators are accelerating the evolution of language
Inside The Viral Words That Make You Click - Etymology Nerd - #1086
As digital platforms accelerate linguistic shifts, understanding how algorithms shape our speech is crucial to recognizing modern manipulation.
- 1Algorithmic constraints and short-form video platforms are accelerating the creation and spread of viral slang.
- 2Content creators rely on specific vocal techniques like uptalk and down talk to maximize audience retention.
- 3The rapid evolution of internet language opens new pathways for top-down linguistic manipulation by bad actors.
Don't miss
Adam Aleksic breaks down the mechanics of uptalk and down talk and how creators use these vocal patterns to keep viewers hooked.
The brief
The rapid rise of TikTok and algorithmic feeds is fundamentally accelerating how human language evolves, turning niche subculture signaling into mainstream global slang overnight.
Linguist Adam Aleksic explains how digital creators use specific vocal techniques like uptalk and down talk to manipulate viewer retention and satisfy strict platform algorithms.
This rapid linguistic shift goes beyond harmless internet slang, raising new concerns about the potential for malicious actors to execute top-down linguistic manipulation.
What was said on this episode
48 statements · 9 positive · 21 negative · 3 mixed · 15 neutral
Clip farming will become the future of online distribution.
“it is a realization that clip farming is the future of distribution online.”
Listen at 0:55
The absurdity of “6 7” constitutes its meaning.
“The absurdity of the word is its own definition.”
Listen at 1:26
Oxford’s word-of-the-year choice was intended to provoke controversy.
“They're hoping that it sparks controversy.”
Listen at 1:43
Words themselves are not inherently good or bad.
“There's nothing inherently in a word that's good or bad.”
Listen at 2:17
Algorithmic trends are accelerating modern slang cycles.
“algorithmic trends being perpetuated that push modern slang cycles faster than ever before.”
Listen at 3:20
Different platforms foster distinct linguistic cultures and dialects.
“a platform functions kind of like a house. It is a place where you go to use a certain type of language.”
Listen at 3:53
Language use signals membership in a social cohort.
“Language is a tool of identity, and when you use a word, you are signaling that you're part of this cohort.”
Listen at 5:29
The linguistic founder effect makes later speakers imitate earlier influential speakers.
“there's this thing called the linguistic founder effect, where you kind of follow in the footsteps of people who came before you linguistically.”
Listen at 6:52
Influencer accents are optimized to perform well in recommendation algorithms.
“the accent is optimized for the algorithm.”
Listen at 7:31
Uptalk helps online creators retain audience attention.
“that uptalk, where you kind of lengthen your final vowel, is very good for online hooking.”
Listen at 7:48
Down-talk signals viewers that they can stop watching.
“it's a signal to scroll away.”
Listen at 9:19
Online performance incentives are homogenizing speaking styles.
“what you're describing here, that you've modified your speaking style for what will perform better online is kind of exactly this greater homogenization effect that's happening.”
Listen at 11:54
Nonstandard accents may reduce online virality because viewers expect familiar speech.
“The cost is you might go less viral because the viewer has an expectation that I want you to be speaking the way I expect you to speak.”
Listen at 12:54
Algorithms compress language into recognizable forms while generating new linguistic niches.
“the algorithm, for example, I think is one such bottleneck that it compresses our language into. You have to be speaking in these widely recognizable accents, but then it speciates and creates new environments.”
Listen at 13:28
MrBeast deliberately changes his accent to maximize viewer attention.
“He is very deliberately switching his accent to grab your attention as much as possible.”
Listen at 14:06
Shock and awe can prevent viewers from scrolling away.
“You just, you want the viewer to remain so dumbfounded watching the video that they don't even think to scroll away.”
Listen at 14:42
4chan incubated slang that later diffused into mainstream culture.
“4chan was a linguistic incubator for decades and, well, a decade.”
Listen at 17:20
4chan anonymity creates pressure to demonstrate slang proficiency and group membership.
“There's the anonymity on the platform where users need to demonstrate a shared proficiency in slang to show that they're not a normie.”
Listen at 17:39
TED Talks should prioritize clip farming because traditional formats have declined.
“the way people should be doing TED Talks now is just clip farming, which is the future.”
Listen at 23:26
Online distribution matters more than content quality for reaching audiences.
“distribution matters more than the content itself.”
Listen at 24:02
Rage bait and clickbait spread more readily online than calming content.
“The things that are going to spread are rage, bait and clickbait.”
Listen at 25:47
Wellness influencers often perform an aesthetic version of wellness rather than genuine wellness.
“Wellness influencers are performing the idea of wellness.”
Listen at 26:10
Emojis function linguistically because they communicate meaning between people.
“they absolutely are linguistic in the sense of they carry meaning and they communicate something from one person to another.”
Listen at 41:18
Emoji meanings continually change over time.
“Emoji definitions are also constantly changing.”
Listen at 43:10
Changes in reality cause language to change as people describe new circumstances.
“If our reality changes. Yeah, language will change as a result, because now we have to describe something different.”
Listen at 44:26
Young people, especially middle schoolers, drive much contemporary language change.
“Most of language change, I would say is driven by people like right now, honestly, middle schoolers.”
Listen at 46:13
Perceived top-down pressure can reduce adoption of new words.
“If you feel like someone is pushing a word onto you might actually not want to adopt it”
Listen at 48:00
Algorithms now use virtually every spoken term to classify and distribute videos.
“Every single term now is a search engine optimization term because the algorithm is looking at every single word you use”
Listen at 53:28
Use of “delve” has increased 1,000% since before 2022.
“usage of the word delve has spiked a thousand Percent since before 2022.”
Listen at 55:10
ChatGPT uses “delve” ten times more frequently than ordinary language.
“ChatGPT uses the word Delve 10 times more than regular”
Listen at 55:18
ChatGPT is influencing humans to adopt different language.
“We are now being trained by ChatGPT to use different language.”
Listen at 56:39
ChatGPT’s word choices are influencing human language use and perceptions of reality.
“the fact that we are also proven to be saying those words more after ChatGPT has been proven to say those words more means that its reality is influencing our reality.”
Listen at 1:00:30
Most of the world’s languages are predicted to disappear by century’s end.
“it's predicted that most of them are going to die out by the end of the century.”
Listen at 1:02:18
Language extinction reduces linguistic diversity and constrains expressive possibilities.
“that homogenizes and constrains the creative potential for expression.”
Listen at 1:02:46
Algorithms replicate human social networks faster and across more connected nodes.
“algorithms have created a replication of natural human social networks that operates faster, that connects more nodes than ever before”
Listen at 1:04:48
Algorithmic networks accelerate both idea diffusion and misinformation.
“which means these ideas can spread faster than ever before. What that means is also misinformation can spread faster than ever before.”
Listen at 1:04:58
Active information-warfare campaigns are currently shaping online information.
“there are active, like, information warfare campaigns occurring as we speak.”
Listen at 1:08:16
Human speakers develop alternative words in response to censorship constraints.
“language can truly be constrained. I. I wrote my book on algo speak, on how words emerge in response to censorships.”
Listen at 1:09:07
Broad generational categories such as Gen Z are social constructs rather than real entities.
“I don't think Gen Z exists. I don't think generations exist.”
Listen at 1:11:07
Institutions manufactured Gen Z as a consumer category that people now perform.
“they have manufactured Gen Z as a concept that now we're also performing what it means to be Gen Z”
Listen at 1:11:57
People should consume diverse media, ideas, and perspectives to counter algorithmic bias.
“we should be consuming as many different forms of media as possible and we should be consuming as many different ideas and perspectives as possible”
Listen at 1:17:24
It is difficult to establish that one language is more complex than another.
“I think it's really hard to make a claim about one language is more complex than another or something like that.”
Listen at 1:25:33
Animal communication is not decodable in the same way as human language.
“Animal communication is just, like, not a decodable thing in the way human communication is.”
Listen at 1:26:40
Language builds social bonds beyond merely transferring information.
“language is also this ritualistic bonding thing between people.”
Listen at 1:29:45
Highly information-dense Ithkuil lacks native speakers because human languages serve social connection.
“no native speakers of ITHQL exist because at the end of the day, we are humans using language to connect with other humans.”
Listen at 1:31:42
ChatGPT predicts English-like tokens rather than genuinely speaking English.
“I would be very hesitant to say chatgpt even speaks English, rather than it's just predicting tokens based on a statistical model of what English should be”
Listen at 1:32:15
Tokenization and model processing can lose meaning during language generation.
“A lot of meaning can get lost.”
Listen at 1:33:29
ChatGPT outputs a mathematical representation rather than any individual’s natural speech.
“It's a mathematical representation of speech.”
Listen at 1:34:59
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.