Brainrot or Language Growth? Therapy Speak, Language Change, & Social Media
Nexus finally put our phones down to interview two campus experts regarding how social media changes language and our growing use of therapy speak in the writer’s room, you gaslighting toxically masculine covert narcissist.
Toby met with Andreea Calude, a UoW linguist and an author who published “The Linguistics of Social Media” two years ago. Between writing her new book, TVNZ, Seven Sharp, & radio, interviews, Andreea sat down with us & spoke on everything from language learning models (LLMs) to dark web surveillance.
For a psychological angle, Toby sat down with Andre Mason, a UoW lecturer & registered clinical psychologist. Usually busy aiding PhD students, researching, & teaching our second-year clinical psych paper, Andre provided Nexus with valuable insight regarding ‘therapy speak’ & social media. Andre has also studied which psychological factors are likely to lead to vaping, oddly enough being a Nexus Writer was not a main predictor.
Toby: Is social media the largest change language has experienced since the printing press?
Andreea: Yeah, it’s hard to tell because quantifying language change is just notoriously difficult, but we think that language on social media changes possibly faster than in other realms because of the continued engagement that takes place online. Geography and time are no longer a barrier. The travel of language is instantaneous; while you sleep someone else is waking up and reading and re-sharing content. Because of that, it enables a faster pace of change.
Toby: As someone who’s had your work unwillingly taken by Meta [to train AI models], how do you feel regarding the injustice that if you took their intellectual property, you would get sued?
Andreea: Yeah, obviously that’s not great. like I said I’m an optimist by default so one of my thoughts was ‘well maybe the upside of this is the data that is being trained on is better’. If I force them to take it out of their data pool then what is going to be left there, it’s all going to be hate comments and misinformation.
I mean sure, it’d be nice if they paid me for that work because they obviously make enough money. But I’m philosophical about it, it’s not great, no. One of the sad things is that you can’t really make money out of writing and I think that’s a consequence of the fact that people read all sorts of stuff online instead and possibly because books are so expensive but that’s another conversation. The music industry is the same, right? It’s very, very difficult. And that’s a real shame because creative industries, we need to preserve them.
Toby: With algorithms compartmentalising users based around interest, does this encourage new jargon to form and dissolve demographic barriers?
Andreea: I mean, jargon is always going to form, it’s not just a social media phenomenon. In sociolinguistic theory, we talk of a community of practice, which is a group that basically shares a common goal, in which members know each other and share a repertoire for communicating to one another (a community of practice necessarily implies these three things). And it doesn’t have to be a positive goal either. Gang work is a community of practice as members have their own repertoire.
Toby: Especially with something like crime there’s a covertness to it.
Andreea: Yeah. You know, some years ago I was chatting to New Zealand police about the dark web, and so one of the really tricky parts of online language that it’s hard for state customs and intelligence officers is to track the words that are used to talk about some of the covert activities taking place there because there is a pressure to use words that the rest of the people in the group will know but at the same time that are hidden and covert enough that the people enforcing the law don’t understand. The online world mimics a lot of what happens in real life.
Toby: With the globalising effect of social media, you would expect languages, or at least dialects of a given language, to merge. Has this happened and if so (or not) why?
Andreea: In a nutshell, yes there are spaces where minority and minoritised languages/dialects/codes can flourish but there are also many where English dominates. Ultimately, it depends on speakers and what they choose to do and use.
Like in our case, there’s Māori Facebook communities where basically the language of that group is Māori. There are spaces like that and there is no merging because the whole point is to kind of allow the language or the dialect to thrive uninhibited from these other pressures. Yet simultaneously there is this kind of global English that is circulating. As academics we know this only too well because with academic language there is pressure to publish in English if you want anyone to cite you.
“I was chatting to New Zealand police about the dark web”
Toby: Gen Z & Millennials seem to use more ‘therapy speak’ than other generations, is social media a cause of this? Or is this because more people attend therapy?
Andre: This is likely due to massive societal shifts about the acceptance and awareness of mental health over the last 10-20 years. Our understanding of what good mental health looks like and the ways that poor mental health impacts our entire lives (e.g., physical health, financial outcomes, lifespan) has dramatically increased. I think a big component of this has also been the number of high-profile individuals who have openly spoken about their own struggles with mental health. Social media has certainly provided a platform for people to share experiences about mental health (we see this a lot with TikTok, Instagram, and Reddit) but I would see this more as a vehicle that reflects these more general changes. The other side of this coin is the amount of misinformation there is about mental health on social media. Increasingly, we’re seeing individuals wanting ADHD or BPD assessments based on what they’ve seen on TikTok. Again, there are pros and cons to everything.
Toby: Can students discuss mental health online in safe ways?
Andre: Yes, there is some fantastic work coming out of the #CHATSAFE initiative led by researchers from Auckland and Melbourne. Social media is here to stay so one of the key questions is exactly this… how do we help young people use online communication safely?
Toby: How does discussion around mental health differ between platforms?
Andre: Greatly but this is largely to do with the social media style (videos, text, photos etc) and the dominant demographic using the platform.
Chomsky, LLMS, & Computational Linguistics
Andreea: I mean, a lot of his [Chomsky’s] theory has been incredibly influential for computational linguistics. All of these applications are based on his approach to language. Because he took such a mathematical approach to language ‘how do we capture what is allowed and rule out what is ungrammatical?’. Cognitive linguistics takes a brain centred approach, ‘what can we learn about the mind from language?’ Whereas he asks, ‘how can we get machines to do language?’ They both have extremely useful merits, but in different ways.
Linguistics is in the science faculty at a lot of universities, not here though, and actually the government sees it as STEM.
A myth is that the LLMs understand language and they don’t. They don’t understand nothing at all. They’re just stats on steroids. There’s a paper called Stochastic Parrots by Emily Bender and her colleagues. It’s a really famous paper in which she talks about some of the pitfalls of AI.
I think the trouble is the tech industry has outrun the legal profession, and so we don’t have good policies in place and this is where government needs to step up and regulate, regulate, regulate, regulate.
Toby: Is Chomsky to blame for LLMs?
Andreea: Oh, gosh. I don’t know that he was directly involved in any of this. But, of course, all the theories of language that we’ve all contributed to in some ways have brought us here. But I think the real problem is a lack of engagement and imagination in seeing the impact down the track, you know? It’s a bit like blaming, I don’t know, Einstein for the nuclear bomb or something.