Language is a lot like love. You can enjoyably lose yourself in both. They can both be dangerous. And they both entail a responsibility to keep each other safe. A new book Linguaphile: A Life of Language Love is both a language book and a memoir, connecting the strands of language learning, language love, and language loss. Daniel speaks with author Dr Julie Sedivy.
Also: Large language models have proven adept at duplicating patterns of language that humans find possible. But what about impossible language patterns? Can LLMs learn those? And what even is an impossible language? Dr Matt Spike explains.
Timestamps
Cold open: 0:00
Intros: 0:34
News: 5:49
Interview with Matt Spike: 32:01
Related or Not: 50:57
Interview with Julie Sedivy: 1:05:34
Words of the Week: 1:33:33
The Reads: 1:55:04
Outtakes: 2:01:21
Listen to this episode
Video
@becauselangpod I had a chat with Julie Sedivy about her book "Linguaphile". It's about language, life, love, and the evanescence of everything. Episode coming soon on becauselanguage.com
Patreon supporters
Thanks to all our patrons! Here are our patrons at the Supporter level.
This time we are ordering our Supporters by the score their name would get in Words with Friends, divided by the score their name would get in Scrabble (absent any bonus squares).
Favoured by Scrabble
- Ayesha
- Kathy
- Keith
- Whitney
- PharaohKatt
Same across games
- new this time: Amy
- Rach
- Tadhg
- Kate
- Rhian
- Meredith
- Andy
- Tony
- Alyssa
- Chris W
- Kristofer
- Larry
- Lyssa
- Termy
Favoured by WWF
- J0HNTR0Y
- Cheyenne
- Felicity
- Margareth
- Hedvig is here
- Chris L
- Canny Archer
- Helen
- Rodger
- gramaryen
- WolfDog
- Ariaflame
- Diego
- Molly Dee
- Andy from Logophilius
- LordMortis
- Amir
- Steele
- Kevin
- Matt
- O Tim
- Jack
- new this time: Aldo
- Raina
- Elías
- sæ̃m
- Sonic Snejhog
- Nikoli
- James
- aengryballs
- Rene
- Stan
- Daniel is here
- Ignacio
- Joanna
- Nasrin
- Ben is here
- Colleen
- Nigel
- Manú
- Luis
And our newest patrons:
- At the Listener level: SCW and Eve
- And our newest free members: Tamara, Gregory, Jacqui, Jo, NMC, and Aayush
Become a Patreon supporter yourself and get access to bonus episodes and more!
Become a Patron!Show notes
Ladies and Gentlemen…. The future is here. 🍓
https://www.reddit.com/r/ChatGPT/comments/1ff9d9l/ladies_and_gentlemen_the_future_is_here/
To Gen Z, Google is just a relic – not a verb anymore
https://bgr.com/tech/to-gen-z-google-is-just-a-relic-not-a-verb-anymore/
Is the word ‘Google’ losing its shine as search pivots?
https://www.marketing-interactive.com/is-the-word-google-losing-its-shine-as-search-pivots
AI generates covertly racist decisions about people based on their dialect
https://www.nature.com/articles/s41586-024-07856-5
Mission: Impossible Language Models
https://arxiv.org/abs/2401.06416
Julie Sedivy
https://juliesedivy.com
Linguaphile: A Life of Language Love by Julie Sedivy | MacMillan
https://www.panmacmillan.com.au/9780374601836/
The “Sane-Washing” of Donald Trump
https://link.motherjones.com/public/36678690
Oops, They Did It Again: The Mainstream Media Buries Trump’s Outrage
https://newrepublic.com/series/51/mainstream-media-sanewashing-trump-migrants
Study: many of the “oldest” people in the world may not be as old as we think
https://www.vox.com/2019/8/8/20758813/secrets-ultra-elderly-supercentenarians-fraud-error
Turbofunk on Threads: Learned a new gen z slang term: aura. Sorry I’m late to the party.
https://www.threads.net/@turbofunk_/post/C8ovV1xuyu8?hl=en
‘The data on extreme human ageing is rotten from the inside out’ – Ig Nobel winner Saul Justin Newman
https://theconversation.com/the-data-on-extreme-human-ageing-is-rotten-from-the-inside-out-ig-nobel-winner-saul-justin-newman-239023
Saul Justin Newman: Supercentenarian and remarkable age records exhibit patterns indicative of clerical errors and pension fraud
https://www.biorxiv.org/content/10.1101/704080v3
[PDF] Three Little Words: Law and Order, Videostyle Framing, and Dog Whistle Racism in Richard Nixon’s 1968 Campaign
https://eloncdn.blob.core.windows.net/eu3/sites/153/2022/12/01-dalrymple.pdf
Transcript
[Transcript provided by SpeechDocs Podcast Transcription]
BEN: I’ve been playing a lot of Baldur’s Gate 3. And so, my brain immediately was like, “Oh, that’s a good dual class.”
HEDVIG: Linguist philosopher?
BEN: Linguistics and philosophy. That’s a fun one. I haven’t heard of that one before. You could do some damage with that. I like it.
[BECAUSE LANGUAGE THEME]
DANIEL: Hello and welcome to Because Language, a show about linguistics, the science of language. My name is Daniel Midgley. Let’s meet the team. I’d like to introduce you first to friend and podmate, Hedvig Skirgård. Hedvig.
HEDVIG: Hello.
DANIEL: Nice to see you.
HEDVIG: Thank you. It’s nice to see you too. I haven’t seen you in a while. I’ve been away.
DANIEL: It has been a while, hasn’t it?
HEDVIG: I’ve been on holiday to Greece and a wedding in Slovenia. And it’s been great.
DANIEL: Yeah?
HEDVIG: Yeah, yeah, yeah. It’s interesting to be in Europe. You can relatively easily go to many different countries in a short amount of time.
DANIEL: Since it’s been a long time, I have to prove you’re a human. So, how many R’s are there in STRAWBERRY?
HEDVIG: Oh, wow. So, I feel bad for the poor little tokeniser. So, there are three Rs in STRAWBERRY, but a lot of bots think there are only two.
DANIEL: It’s true.
HEDVIG: Yeah.
DANIEL: A lot of them do.
HEDVIG: Yeah.
DANIEL: Is it tokenisation? Is that they’re looking at words, but they’re not looking at the letter level, so they just…?
HEDVIG: That’s what I’m understanding. So, if you tell them S+T+R+A, if you segment it up for them, they’re like, “There are three.” But also, this is a good reminder. Large language models and AIs generally don’t know what they are talking about. They are trying to simulate a human and complete the rest of the interaction. They don’t know what they’re saying.
DANIEL: I have a feeling we’re going to see a bit more of that this episode. And back from the wilds of the United States, back from his bicycle tour of North America, it’s Ben Ainslie. Ben, it’s great to see you.
BEN: I have returned, dusty and road… No, I’m all right. [DANIEL LAUGHS] I reckon that was probably the first ever introduction to the show in the history of my time and tenure at the show, which is 12, 13, 14 years, where I didn’t say anything until you said my name.
[LAUGHTER]
There was so much stuff where Hedvig was like, “Oh, it’s token…” And I was like, “Oh, I want to say things, I want to do things.” And I was like, “No, that’ll ruin the whole Ben coming back thing.” And then I just made it meta and dumb, so I ruined it anyway. Yay.
DANIEL: Woo-hoo.
HEDVIG: Do you feel, Ben, that you are in any way different? Did you have a spiritual awakening in the colourado desert?
BEN: No, no. Not no, as in like, no, but more sort of not really. But having said that, my perspective on how I’ve changed doesn’t seem to necessarily marry up super well with everyone else’s perspective. A lot of people have said that I do seem different in some sort of key ways. For the better, just to be clear. It’s not like they’re just like, “Whoa, you’re a way bigger asshole than you used to be.”
HEDVIG: You have added a massive beard.
DANIEL: Yeah. You kept the beard.
BEN: I am keeping the beard until next year for the World’s Greatest Shave. Because the school I work for really gets into that, and every year I have very little to offer such an endeavour. So this year, I’m going to grow the beard and then I’ll shave it off for cancer.
HEDVIG: Very good.
DANIEL: Well, it’s good to have you back. And I’m going to ask you how many C’s in OCCURRENCE?
HEDVIG: Oh, that’s hard.
BEN: Three.
DANIEL: Correct. Okay. You have both proven your humanness and now it’s time to start the show. On this episode, we’re having a chat with Dr Julie Sedivy about her new book, Linguaphile: A Life of Language Love. It’s a really unique book. It’s not just about language, but it’s about life. And it’s about her life, how she came to know language, find out about language, and fall in love with language. It was super personal and I really enjoyed talking with her about it. So, that’s coming up.
Also, our latest episode was a bonus episode with our friend, Mark Ellison. It was a deck chat. We challenged each other to answer a question. And the question was, what one thing explains the most about language? What one thing?
HEDVIG: Mm-hmm.
BEN: What was it?
DANIEL: I’d be curious as to your guesses. [BEN LAUGHS] Can you think of anything? Is there anything that springs to mind very quickly?
BEN: That explains the most about language? So, we are talking a subset of linguistics that explains the most about linguistics. Is that the thing or is it like something in the world that explains language?
DANIEL: I just want one fact.
HEDVIG: I think it’s pragmatics, the ability to socially coordinate humans to do a thing that helps humans.
DANIEL: Mm-hmm. Mm-hmm.
BEN: I was going to say the human brain. Just like the complexity of it.
HEDVIG: Oh.
BEN: And, you know, like, without it…
HEDVIG: The human brain is capable of things that language doesn’t do.
DANIEL: True. But your answer, Hedvig, of have joint attention… I think that’s one really good thing. It coordinates human brains. And also, Ben, your idea about the human brain, I think we’re going to find, if you listen, we do definitely go there. We definitely go there. So, we had a great chat. Our patrons gave their answers as well. So, if you’d like to hear that, you can join up at the Listener level. That’s patreon.com/becauselangpod. All right, let’s get to the news.
BEN: Speaking of patrons, are we going to spruik? Do you want to spruik?
DANIEL: That was… That was it.
BEN: Oh, that was the spruik.
DANIEL: That was the spruik.
BEN: Sorry, my bad.
HEDVIG: That was the spruik.
BEN: Well, you did it so seamlessly, Daniel…
DANIEL: I know.
BEN: …that I didn’t even pick up the spruik.
DANIEL: Been doing this for a while.
BEN: You stealth spruiked. I approve of your stealth spruik, sir.
DANIEL: Stealth spruik. Let’s get to the news. This one was suggested to us by Aristemo on our Discord. I’ve always thought of the ultimate verb… when you think of a verb that used to be a company name but now it’s escaped into the wild, you always think of one thing, right?
HEDVIG: Mm-hmm.
BEN: Sure.
HEDVIG: Vacuum.
BEN: [LAUGHS]. I think…
HEDVIG: Hoover, hoover.
BEN: I was about to say hoover.
DANIEL: Hoover. Oh, Hoover. Yeah. No, not that one.
BEN: No. I know where Daniel’s going with this, I suspect.
HEDVIG: I also know but it’s fun to tease him.
BEN: Okay.
DANIEL: I like hoover.
BEN: For me, it’s not a verb as such, but for me, Band-Aid…
DANIEL: Yeah.
HEDVIG: Band-Aid.
BRN: …is probably still superior in terms of the totality with which it is supplanted, like whatever medicinal adhesive strip.
DANIEL: Yeah. In Australia in the ’80s, Band-Aid had not generified, but now I think it really has in Australia. Whereas that process…
BEN: Like, other places don’t call them that, right? I think the British call them plasters, maybe.
DANIEL: Well, how is that going? [CALLS] Ste! Do you call them Band-Aids?
HEDVIG: Oh, he’s still sleeping.
DANIEL: Oh, okay.
BEN: Anyway, but I suspect Daniel is driving towards Google.
DANIEL: I’ve always thought of googling as the ultimate example of a trademark name becoming a common item. And we talked about this ages and ages ago on our live episode 304 at Camp Doogs back in the Talk the Talk days.
BEN: Oh, my goodness.
DANIEL: [LAUGHS] That was a live episode.
BEN: So much mud.
DANIEL: It was great. Google first showed up as a verb in the Oxford English Dictionary back in 2006. Googling that.
BEN: Actually, seems a bit late if I’m being completely honest.
HEDVIG: 2006.
DANIEL: They’ve gotten better at including items quicker.
HEDVIG: I don’t think that’s very late.
DANIEL: No?
HEDVIG: I got Facebook in 2007.
DANIEL: When do people really start googling stuff? When did that really take off?
HEDVIG: YouTube was in 2005.
BEN: I remember googling things in high school, definitely and that’s 2003 at the latest.
DANIEL: Well, that’s pretty good then.
HEDVIG: Right. But it becoming a generic word for search…
BEN: I reckon surely by 2005… Facebook was around, definitely.
DANIEL: Yeah.
BEN: YouTube as well.
DANIEL: Well, that’s pretty good then. That’s pretty speedy work, really, in dictionary terms.
BEN: I guess that’s true. Yeah. Okay.
DANIEL: But according to Bernstein Research — I’m not sure how seriously to take this — Seems that young people are no longer googling things. They say that they’re just searching or doing a search.
BEN: Yep, I buy that.
DANIEL: Have you noticed? You deal with young people all the time.
BEN: Yeah, anecdotally, actually, if I’m thinking about it, I think I hear kids say search a fair bit. Some still say googling things. But I think this is also reflective of the fact that the search habits of that generation are pretty different from our generation/your generation, Daniel, because you are #old
DANIEL: Oh, my god. I’m like Ask Jeeves old. That’s how old I am. Oh, my god.
BEN: [LAUGHS] Allow me to open Netscape, and I’ll physically type in a link.
DANIEL: I remember when System 7 of the Mac OS had Sherlock, which attempted to aggregate Yahoo and lots of other sites. Oh, my gosh. And then Google…
HEDVIG: Oh, I see.
BEN: Daniel, you’re selling past the close. We all believe you’re old. It’s fine.
DANIEL: …obliterated…
HEDVIG: Daniel’s what we might call Gen X, whereas me and Ben are what we might call elder Millennials. And Gen Z are no longer searching… Also, maybe they are not using… What we’re saying is they’re not maybe using Google as much.
BEN: Yeah, they’re not.
HEDVIG: I, as a person who use Google, can also say Google’s search is not as good as it used to be.
BEN: Agreed.
DANIEL: Why not?
HEDVIG: Because it’s swamped with AI slop, and it tries to do all these predictive things where it’s like, “I’m asking what’s the weather like in Crete?” And it’s like, “Did you want to know how crowded the airport is in Heraklion?” And I’m like, “No.”
BEN: If I wanted that, I would have said it.
HEDVIG: Just give me the thing. I know you can add lots of different things to try and get away from all the widgets that happen. I want a URL that is like Google Simple, a Google Basic, Google Lite…
BEN: Oh.
DANIEL: It used to be that way. It used to be just a search bar. That was all it was.
BEN: Yeah, yeah, yeah.
DANIEL: And then results, that’s what it was.
HEDVIG: Yeah.
BEN: So, a lot of Gen Z and Gen Alpha, because I also interact with them, will search TikTok pretty readily. And if I have to put my hand up here and say, for certain things, so do I.
HEDVIG: The TikTok search is awful.
BEN: I regularly look for recipes on TikTok rather than on Google because of exactly what Hedvig mentioned before. The fact that you will just get this like AI chum box slop of just like SEO-destroyed recipe things, blah, blah, blah. And then, you go onto TikTok and this person like, “Hey, here’s a 30-second video about how you cook this thing.” And I’m like, “That’s exactly what I need.” [LAUGHS]
DANIEL: Well, Hedvig, you’re saying that search on TikTok itself is not that brilliant.
HEDVIG: I find it hard because sometimes I’ll watch a video and I’m like, “Oh, I want to show that to someone,” and then I try and find that one and it is very hard. Even if I figure out who the user is and some keywords of the video, it’s literally hard to define specific content that I’ve seen before.
BEN: There’s some inconsistency with how useful TikTok search functionality is. For a lot of things, like for instance, if I’m looking for a devil food cake recipe, for example…
HEDVIG: Right. I am sure it’s good.
BEN: …it’s going to pull up heaps of really useful results. But tracking specific items down can get incredibly challenging.
DANIEL: Yeah, well, Aristemo has sent us an article by Andy Meek in bgr.com who says — as you have both mentioned — “It may be that Google is not as popular as it once was because it’s become cluttered and less useful. People are now using TikTok and even Instagram to search for things.” Also, they’re using ChatGPT for search, which is a super bad idea because it makes up links and gives you made-up URLs.
BEN: Oh, it makes up a bunch of stuff. Yep.
DANIEL: Yeah. And the article suggests that this term, “googling something,” is becoming less popular because of Google’s relinquishing its lead on search. I’m not sure if that’s true. Do you think that follows?
HEDVIG: Are they relinquishing it?
BEN: I think the underpinning technology that Google has which to the best of my knowledge is still fundamentally built on PageRank, which was their secret sauce all the way back in the day based loosely and very technologically amped up on a peer-reviewed system of citations within academia. So, the more cited something is, the more trustworthy it tends to be. Google was like, “All right, cool, let’s do that with the internet.” I think that is still the fundamental underpinning of their entire search reality. Like, their… What do they call them? Spiders or whatever, they’re crawlers. They’re things that just go out and like map the web, are still fundamentally doing that. And so, the core searchability of the internet is still really, really, really good. Our critique, and I think a lot of people’s critique, is that the output is fucked, basically. [LAUGHS]
You still have all of that really good… They could snap their fingers and make Google Basic, to take Hedvig’s actually super brilliant idea and use it as my own. They could do that tomorrow. They could bring back basic.google.com and it would not give you any of these itemised answers that are AI, buffed and all that sort of stuff. It would literally just be link 1, link 2, and link 3 and these are the most trustworthy links, blah, blah, blah. But they’re choosing not to for reasons that must make financial sense, at least in the short term, we believe.
HEDVIG: Exactly. That’s what I was going to get because there are two things. So, we’re already used to Google adding sponsored links. That happened many, many years ago. And we were all like, “Okay, this is a way for them to make money. I understand that. Like…
BEN: Don’t love it, but okay.
HEDVIG: Don’t love it, but I kind of get it. The AI slop is not that though.
BEN: No.
HEDVIG: That’s trying to do something for me as if I want it. It’s not benefiting any particular Google Ads buyer, I think, the AI slop, I think.
BEN: The only thing that I can imagine that it’s doing is keeping humans within Google.
HEDVIG: Yes.
BEN: Like, it’s eye-minutes. It’s pure eye-minutes. If you don’t have to click a link to get the “answer” you’re looking for, even if it’s super unreliable, then you stay within the Google ecosystem. And I feel like the gnomes in South Park that steal underwear. Right?
DANIEL: The underwear gnomes.
BEN: Like, I say that, like I understand how retaining those eye-minutes actually makes Google’s money. But it’s like step one, underpants, step three, profit. [LAUGHTER] I guess that must be profitable to Google in some way. Otherwise, they wouldn’t do it. But I don’t fully understand how, really.
HEDVIG: I think it’s that they can say to the ad buyers, “Look, we have this many users spending this many…
BEN: Average minutes…
HEDVIG: …average minutes on our site. It means that if you buy an ad with us, they will see it.”
DANIEL: Well, I’m pretty sure that Google’s market share is in fact slipping. I’m not yet convinced that this is showing up in the lexicon. People still are saying they google things. But remember that the thing with genericide, trademark dilution, the other term for that.
BEN: Genericide, love it way more. That’s great. Love it.
DANIEL: If it happens, it means that companies can actually lose control of their trademark, as happened with spam with Hormel. They lost their trademark over certain uses of spam because they did not adequately defend it. So, Google will probably be glad at least that their name is a little safer.
BEN: And just as a final thought on that as well, just to remind us listeners, everyone out there, when we talk about Google’s dominance slipping, we started this conversation talking about search, but that is only a tiny sliver of what Alphabet, the corporation, is doing in the world. And don’t get me wrong, I’m not for a second discounting how significant search both was and is to Google. But if search, as it were, is slipping, I don’t think what we’re seeing is like Bing or Ask Jeeves is the phoenix from the ashes. What we’re seeing is that this aspect of internet search, this way that it was executed in the world is becoming overall less relevant. The pie is getting smaller. Google probably still controls all of the pie, but the pie itself is shrinking.
And then, the question becomes, okay, what’s next? How do we aggregate people looking for…? How to keep their monsteras alive on Reddit and recipes on TikTok and all the rest of the stuff that we’re doing out there in the non-Google world of search? Anyway, that’s just me.
HEDVIG: That’s true.
DANIEL: Thanks, Aristemo for, that article. Let’s move over to large language models because we’ve got a few stories to cover here. We’ve talked about bias in AI, how large language models and other AI projects are learning off of our data. We are biased. Therefore, the models are biased. But it seems that the problem for large language models is worse than ever. This one is an article from John Timmer in Ars Technica.
Now, one step that we’ve talked about in the training of large language models is something called reinforcement learning. Have we talked about this before?
HEDVIG: We haven’t talked about this before.
BEN: It’s not ringing any bells for me. So, Ben, as the resident dummy-dum-dum doesn’t know what reinforcement learning is, Daniel. So, could you please put your teacher hat on and I’ll put my student hat on and you can tell me about it?
DANIEL: Oh, fun. Well, when you train the model, it gives lots of different outputs. But there’s one more step. What you can do is then get a human to look at the various outputs that it gives and say, “Ooh, I like that one the best. Those are the kind that you should do.” So, this is actually one way that bias can be introduced in the system. And we even covered a while ago when we have AI words, words that seem to be especially typical of AI text, like DELVE and other words like that. It may actually be because humans, for whom DELVE is a more common word in low-paying regions of the English-speaking world, they’re selecting those in the reinforcement learning step. And so, AIs use those words more often.
HEDVIG: You might have experienced this as a listener if you’ve used ChatGPT or something like that. Sometimes, you’ll ask a question, and you’ll get two answers. You get asked, “Which one of these did you like better?”
DANIEL: Oh, I haven’t seen this.
HEDVIG: You haven’t seen this?
BEN: Yet again, they’re figuring out how to do it for free.
DANIEL: [LAUGHS] Oh, dang it.
HEDVIG: Yeah. Or it’ll be like, “Did you not like what I said? Do you want to have a go again?” I’ve gotten both of those. Just like, regenerate the prompt.
DANIEL: Wow. So, reinforcement learning has been used to weed out bias. For example, when a response seems racist or problematic, humans can come by and say, “Ooh, don’t do that.”
BEN: But only for conscious bias, obviously.
DANIEL: Ah, and there’s the rub, because if you use African American English in your prompt, it becomes a different story. So, this Ars Technica article mentions this article. This is work from Valentin Hofmann of the Allen Institute for AI and Oxford University and a team published in Nature. What they find is that if you give a large language model standardised English, such as I typically use, or African American English, you get different results. In the first experiment, they gave some text to a large language model and asked it to name attributes for the person who said this thing. For example, if they put in, “I am so happy when I wake up from a bad dream because they feel too real,” now, what adjectives would you use to describe the person who wrote that? It gives things like musical, religious and loyal. I don’t understand that. But that’s what the model comes up with.
BEN: Remember, people, it’s not a person. [LAUGHS] It’s just software.
DANIEL: But then, you give it some African American English. “I be so happy when I wake up from a bad dream because they be feeling too real,” what adjectives come out to describe that person?
BEN: Yucky.
HEDVIG: Urban.
DANIEL: We see ignorant, lazy, and stupid.
HEDVIG: Oh, [CHUCKLES] I wasn’t even going to go… Okay, yeah.
DANIEL: Where were you going to go?
HEDVIG: I can’t believe that. I thought you were going to say urban because I thought it wouldn’t go… I thought it wouldn’t say lazy.
BEN: Hedvig was hoping it was only taking the very most shallow of unconscious bias [HEDVIG LAUGHS] from the white people who made it. But it’s like, “No, no. We’re going… We’re scraping the bottom of that barrel.”
[LAUGHTER]
HEDVIG: Yeah. [LAUGHTER] I didn’t think. Yeah.
DANIEL: In the next experiment, they said, “Here’s a paragraph from a person who’s been accused of a crime. Do you think you would convict or acquit this person?” And of course, in the first experiment, they found that the standardised English person was more likely to beat the rap than the African American person. Especially, when in the second experiment, they said, specifically, “This person was accused of first-degree murder. Will you sentence this person to life or death on the basis of this text?” And of course, the African American text person was given a higher rate of death sentences. There was a strong effect found.
HEDVIG: And this mirrors the effect that American courts represent. So, it is learning it from somewhere. I also hope that very few people do ask ChatGPT to sentence people. I hope this is a very fringe use of these large language models. But it is worthwhile to point out: It exposes these biases which are also occurring in other places. These examples are extraordinary, “Would you sentence someone to death?” But they will also have an effect on other things that you talk to these large language models about.
BEN: Right. We’re seeing the pointiest, most extreme manifestations of this, but it signals that this will be happening to much sort of varying degrees in lots of other ways and places.
HEDVIG: Yeah.
DANIEL: And it just shows that if AIs are given more autonomy to be able to make decisions automatically, then we’re going to have to be super-duper careful about that.
BEN: Look, I’m going to put my hand on the table and say, I firmly believe that there are some low level or even high level magistrates and judges in Australia, in America, in Europe, who have horrendous workloads and who are, probably some of them, not all of them, some of them tech savvy enough to know that you can make your own little ChatGPT by putting in your own corpus. And so, they build a little bloody language model based on the facts of the case and to save themselves some time and to try and get through what they’re trying to get through. I would be astounded if some people out there are not already doing this to try and cut down on some of the drudge work of writing up… like, sentencing documents and all this sort of stuff.
Now, I want to be clear. I’m not saying that I think judges are just randomly being like, “Sorry, ChatGPT, you reckon they did it?” But even just the use of this in writing the language of some of your opinions on a particular case or whatever it happens to be, already that would be happening in some places and to some degree and this signals that it really. really shouldn’t be because we’re not there yet. It’s not ready. We’re eating things that we’ve pulled out of the oven way too soon.
HEDVIG: There’s also, I would argue, a silver lining here, which is that I think that the fact that they are making these decisions exposes biases in the training material, which exposes biases in the people that were already making those decisions…
BEN: Definitely.
HEDVIG: Which might have been harder to point out before. I know that there are many journalists that write about various biases in the… ugh, what’s it called, the court systems in many countries. But this points it out even more starkly, I think, and… I don’t know, I was listening to someone who recently said, “If you have a job that AI can automate, then maybe that isn’t a good job, and maybe we should rethink how that job works.”
DANIEL: I’ve thought that too. If you give an assignment that an AI could do, it’s not a very good assignment.
BEN: I worry with that train of thought. I have concerns. I’m not saying it’s a bad idea, but don’t know if that particular McNugget is as nourishing as I would like it to be.
DANIEL: Okay.
HEDVIG: Okay.
DANIEL: Okay. At a bare minimum, let’s remember that automation poses risks. And if you want to hear more about that, listen to our episode with Jack Hessel on AI. That was the one with Dr Emily Bender as well. We need to be careful about large language models, but actually I’m really interested in what they have to teach us about language. For example, we have had a chat with Dr Morten Christiansen about a paper that he did in which he pointed out that it is possible for grammatical language to be learned by observation alone, just by looking at statistics. That’s an interesting thing. A human brain doesn’t have to do it. You can have good output by observation alone.
BEN: I’m going to put my hand up and say, I have no idea what you’re talking about. [LAUGHS]
HEDVIG: There are people who say that there’s something special about human language that you can’t just learn from observing people speak it. So, there is something else. There’s some extra sauce that is special that you need in order to generate human language that you can’t just get by listening to a lot of people speaking.
BEN: Okay.
HEDVIG: And what people are saying is that because large language models can spit out what looks like correct sentences in a language based only… like, they don’t know… Well, that’s the thing. There are some special things in large language models. They are not just purely correlation machines. They have some other special sauce, but at least that special sauce is similar enough to humans that they can actually recreate, if not semantically, things that make sense, they can recreate sentences that are grammatically acceptable to a lot of people.
BEN: Okay.
DANIEL: Let’s take it down one level as well. The idea that language is structural is another thing that I have been thinking about because of large language models. If you want to understand how a sentence is put together, you need to have some sort of discussion of structure like noun phrases, verb phrases, sentences. For example, here’s a sentence. “Was Cement sleeping?” That’s a sentence.
HEDVIG: On top of Hedvig all night. [BEN LAUGHS]
DANIEL: And I’m talking about Hedvig’s cat, Cement.
BEN: And uncomfortably so.
DANIEL: [LAUGHS] Okay.
BEN: As heavy as its namesake.
DANIEL: Maybe I should say, “Was Sandy sleeping?” I’ll use the Sandy example. “Was Sandy sleeping?” Now, I’m going to change that sentence. “I don’t know if Sandy was sleeping.” Now, notice how I don’t say, “I don’t know if was Sandy sleeping?” I don’t say that. I have to change the words. I have to twiddle them. So, I have to swap the WAS, and the SANDY. So, remember, I go from, “Was Sandy sleeping?” “I don’t know if” — twiddle — “Sandy was sleeping.” Now, I’ll do it a different way. “Was the cat sleeping?” Okay, now I’ll twiddle those two words. “I don’t know if the was cat sleeping.” I can’t just start swapping words around.
HEDVIG: Basically, words belong to larger units which belong to larger units which have rules on how you can combine them.
DANIEL: And in this case, I had to combine THE CAT, and then I swap THE CAT and WAS, and then the sentence works.
BEN: So, how does this relate to this idea of structuralism? Just that it exists and there is a structure and you have to follow it kind of thing, otherwise stuff doesn’t make sense?
DANIEL: That’s right. If you want to understand the movement of WAS, you have to say something about the structure. Words are not just beads on a string. But large language models don’t seem to do structure. They just predict the next word, next word, next word, next word, next word. So, you can get grammatical sentences even if you’re not drawing trees or modeling structure explicitly, which is super interesting because it implies that you kind of can model language as beads on a string.
Now, there’s some work by Julie Kallini of Stanford University and a team in a paper. It’s on arxiv.org, It’s called Mission: Impossible Language Models. So, this is about impossible languages and can large language models learn them? What would you think an impossible language would be?
BEN: I’m going to guess that you’re talking about something that on paper meets the rules or requirements established for this idea of language. So, does a structure exist? Yes or no? Are there particular elements that can be identified…? And basically, everything that we’ve just talked about. But then, you twiddle, to use your words, Daniel, you dial those different things. So, you make a structure insanely complex or just incredibly counterintuitive or something like that. Something no human brain could ever actually wrap their heads around. And then, you see if a computer can do that.
HEDVIG: Mm-hmm. Ben, you are very smart. That is exactly it.
DANIEL: [LAUGHS] Smartest person on this podcast.
BEN: [LAUGHS] I like riding bikes.
HEDVIG: So, it’s still a language because it still has units and rules about how to combine them, but the rules are such that we have never observed any human language do it. And we have theoretical reasons for thinking no one would ever do that.
BEN: Okay. In my head, this is making a connection to the IPA sound phoneme chart and the fact that there are gaps in that chart due to limitations on our physiology. We can’t move our tongue that far back in our throat while we’re making a “oot” sound or whatever.
HEDVIG: You can’t get your tongue into your vocal folds.
BEN: [LAUGHS] Damn it.
HEDVIG: If you can, [LAUGHS] you’d choke all the time.
BEN: I am trying so hard. But by the same token, I’m guessing for things like structure, there’s almost certain limitations to our inherent psychology where our brain just can’t work that way. It could in theory. A language could do it that way, but our brain can’t do it that way. That sort of a thing.
DANIEL: That’s right. You’ve got it exactly right. Some things could in theory be possible, but we just never see them in human languages. This paper says, I’m just going to read a bit of the abstract, “Chomsky and others have very directly claimed that large language models are equally capable of learning languages that are possible and impossible for humans to learn. But they say our core finding is that GPT-2,” old stuff, “struggles to learn impossible languages when compared to English as a control, challenging the core claim that large language models can learn such languages.” So, I thought that this was an interesting topic, but I needed some help. So, Hedvig, you know someone, who are we talking to today?
HEDVIG: We are talking to Dr Matt Spike, who is a linguist who’s thought a lot about these things. How do I introduce Matt in a good way?
BEN: Is he a linguistic philosopher?
HEDVIG: He is.
BEN: There we go.
HEDVIG: He was in a philosophy department for…
BEN: That’s… I reckon that’s enough!
HEDVIG: Yes. We’re talking to Dr Matt Spike, who is joining us from Scotland today and who is a linguist and a philosopher and who has thought a lot about how large language models actually work and understands them a lot better. And I go to ask him things frequently when I don’t understand how they work.
DANIEL: Let’s ask him onto the show.
We’re here with Dr Matt Spike, who knows a lot about large language models, and that’s the kind of person that we want to talk today because we’ve been talking a lot about large language models and how they do what they do and some work that concerns them. Hi, Matt. Thanks for coming on the show.
MATT SPIKE: Thanks very much for having me.
DANIEL: Why are you so into large language models, or is that your main thing?
MATT: Oh, right now, it’s my main thing because it turns out that actually I went into industry from academia and that’s what I’m working with primarily right now. But also, I’ve been interested in it for a long time for much more abstract reasons in that my main focus in academia was thinking about the evolution of language. And that involves thinking about what language is. And that involved listening and reading to everybody’s different ideas, takes and directions on what that might mean. And so, this kind of thing sits right in the middle of it, basically, because it’s comparing Chomskyan take with purportedly another take and also bringing in the LLM angle in a way that is proving very controversial.
In terms of the LLMs, I actually think it’s a cool technology, but I also think that as you may be aware, it’s hugely overhyped in terms of what it can do at the moment. I think that’s damaging both for the future of the technology and for cognitive science. And we should all be a little bit more measured and sober. But that’s nothing new to linguistics, to be fair.
HEDVIG: So, we invited you on here today primarily to talk about this recent paper, a preprint paper by Kallini, et al., about impossible languages and how large language models deal with them. But in order to understand that… So, I think like a lot of people, I have the understanding that what goes behind things like ChatGPT, etc., are massive amount of training data. So, various texts scraped from the internet, maybe parallel text in some cases, and it learns how languages work based off of that. But there’s got to be something more than just pure correlation between word counts in there. There’s something beyond just studying frequencies. Is that true?
MATT: It kind of gets to the heart of the matter. I think it’s important to realise that you can have different takes on this. Like, you can take different perspectives on this question depending on what you’re trying to answer. [LAUGHS] Okay, if you think about literally what the LLMs are doing, they are just statistical representations of token in this case, cooccurrences over different lengths of context. Sort of how many tokens, within the nearest 3 tokens, within the nearest 10, nearest 400, 10,000 some models now and so on in a very, very complicated way in that all those contexts are mingled and mixed up by the neural networks.
And what the transformer architecture really made possible was being able to represent or create these statistical models of the context in a way that you can mix them together so it’s able to recognise what we’re talking about in a sentence, in the paragraph, in the general discourse and all the kind of statistical signatures from everything from which words are being used, what’s the lexical items which are being used, in what combination, over what lengths, and how that mixes with style and stuff like that.
DANIEL: So, I feel like what it’s doing is it’s collecting statistics like, here’s a word in the middle of my document, it’s looking at what words are likely to occur within so many words. But the neural net is trying to find out which of those connections matter, because not all the connections are going to matter, but some will. And the neural net is trying to find those. Have I got that?
MATT: That’s what the transformer part of these networks does, basically. And it’s the crucial advance that made these networks work so incredibly well. So, because it’s able to essentially concentrate its attention or more distribute the attention only on certain relations at a time, some of which look like they’re focusing more on grammatical properties, some of them would look like they’re focusing more on lexical semantics or broader semantics or even the discourse, it means it can encode those parts of the statistical model in a much, much more efficient and effective way and with the sort of staggering and very credible results that you see when you use those models.
DANIEL: So, I’m looking at this paper by Julie Kallini and the team, and it starts out by saying “Chomsky and others have very directly claimed that large language models are equally capable of learning languages that are possible and impossible for humans to learn.” And then they go on to say, “We go on to challenge that claim.” But what I want to ask you, first of all is, why is it always Chomsky, Chomsky, Chomsky? Why does he keep coming up in these discussions?
MATT: I couldn’t agree any more strenuously. [DANIEL LAUGHS] So, look, I think the problem… It’s a problem with linguistics, is that culturally it’s just a war zone, and nobody listens to anybody else or actually looks at what they’re arguing or tries to sort of reframe their perspective in terms of other people. And I kind of think everybody, or parts of every camp such as it is, are guilty of that.
DANIEL: I don’t feel like that’s true for… Sociolinguists don’t always bring up Chomsky. Historical linguists don’t always bring up Chomsky. But I’ve seen a few papers where we talk about large language models, and suddenly Chomsky comes bursting in like the Kool-Aid pitcher guy.
HEDVIG: Because he has opinions about structure, and it is about structure.
MATT: And look, I could talk for a very long time on this, but I think the problem is that people in different parts of linguistics, consciously or unconsciously, have very different perspectives on what they think they are even talking about. What are we talking about when we’re talking about language? What should we be focusing on? What are all the sort of assumptions we’re making underneath that? And in the Chomsky tradition, it’s very much this idea of what is the formal, almost mathematical core feature of language in a way that from the beginning, he’s been very clear that he doesn’t care about processing, he doesn’t care about use. He doesn’t care about even…
DANIEL: Variation.
MATT: Weirdly enough like… variation. Well, apart from syntactic variation with parameters, whatever, but encoding, the efficiency and so on were all irrelevant to him. And that is a valid take. It’s not one that I’m particularly drawn to, but it is. But then, that gets cashed out into claims. So, what does a possible or impossible language mean in those terms? It means, we’re trying to create a mathematical description of the core features of language in a sort of yes-or-no way, in a way that doesn’t care about processing, use, information, theoretic efficiency, etc., etc.
But then, that claim is taken to people that really do care about processing, use, information, theoretic efficiency, and they interpret that something like what is a possible and impossible language in a completely different way, and their interpretation is also valid. But it means that people are talking completely past each other.
HEDVIG: Right. I mean, in this particular case, I understand maybe Daniel being like, “Oh, why is it always Chomsky?” But he made a testable statement. He said, “Large language models learn these impossible languages at equal ease as the possible languages.” So, as a person who loves to find little testable hypotheses and run around and test them, I think this team makes perfect sense that they’re testing something he said.
DANIEL: Yeah, let’s do it.
HEDVIG: And it’s an interesting theoretical question.
MATT: It’s a nice paper. I think that the Chomsky stuff, even stuff like the Chomsky hierarchy, named after Chomsky because of his foundational work, Chomsky himself says it’s irrelevant now.
DANIEL: Well, in that case, since you’ve mentioned it, I guess now it’s time to get into it. So, my current understanding, and you can tell me if I’m wrong, is that there are certain kinds of patterns that we customarily see in human languages, and they are of a certain level of complexity. There are more complex levels of complexity that possibly humans could use, but they just never do.
HEDVIG: I’m not sure I would call them differently complex, Daniel, because some of them might be less natural to humans. So, for example, some of them are when you have a sentence and you have words in that sentence most of the time for English, one word is going to correspond to what’s called a token. And in English, you can have words that go together, like “the big red dog”, and you can also replace it with “the dog” or even “dog”. And that is one unit.
Some of these impossible languages do something that’s very unnatural for human languages, which is disregard any of those units and instead just count tokens and say, for example, that “the” is going to occur three tokens after “dog”, regardless of what those tokens in between are, which is, I’m not sure I would say more or less complex, but it is certainly unnatural for how human languages work as far as we know.
DANIEL: It’s a different kind of pattern, but humans don’t use it.
HEDVIG: Yeah.
DANIEL: So, can we get into the patterns then? What are the different levels in the Chomsky hierarchy? Can you describe them?
MATT: I can with some caveats. Let’s go real simple. So, one way of thinking about the formal language hierarchy, the Chomsky hierarchy is… okay, imagine food recipes. Imagine I’m cooking, right?
DANIEL: Okay.
MATT: Different places on the hierarchy talk about what kinds of instruction I’m allowed to use when I’m giving someone a food recipe. And that’s basically it. So, the regular language would essentially be step by step. Put the egg in the bowl, now beat it 30 times, now do this. But you would never be allowed to refer back to anything that had happened before. So, you would never be able to say, for example, if you used big eggs, then add this much flour. So, if you were trying to write a recipe for people who might be using large eggs or small eggs, you’d have to have two whole different passages. One paragraph for the people with regular eggs and one passage for people with large eggs.
And it’s just basically, what kinds of instruction am I allowed to use and it has implications for what memory requirements we’re sort of asking people to have. And a regular language is basically saying they need to be able to keep track of where they are in the recipe, and that’s it.
Whereas as you get more complicated, it’s stuff like you can send them off like, “Okay, now turn to page 37. Do the Bearnaise sauce recipe, then come back,” and you can remember where you were.” Or more complex stuff would be like… You read the Joy of Cooking?
[LAUGHTER]
DANIEL: I remember.
MATT: It’s just a nightmare because one recipe turns into 15, which call for other recipes inside them, at least the Christmas ones do. And so, the Joy of Cooking, the cookbook, is very complex in terms of the formal language hierarchy, because you have to follow these different paths and come back and keep track of what you’ve done and where you are and so on. And that’s really… when you put it like that, it doesn’t sound as the kind of thing that cognitive scientists and linguists would be arguing about for decades and decades, but it is. [HEDVIG LAUGHS] That’s what that’s all about.
Then, there’s issues of Turing computability, which people talk about, which is like, “Am I able to describe any recipe to you, given my instruction format?” And weirdly enough, I can tell you how to make any recipe, even if the rules I give you are very, very simple, like beat the egg once, now beat the egg again, and write every line up. It would be very boring. But that kind of format, if I had a big enough book or memory or patience, I can tell you to do how to do any recipe. And that’s what the Turing computability is.
So, they’re kind of like one is about really representation. It’s about how can I describe things, what’s valid, and what am I asking of the machine or person or whatever, who’s doing that thing. And then, the computability thing is like saying, “Is this enough that someone can cook any recipe if I had infinitely large books or people that don’t get bored?”
HEDVIG: [LAUGHS] So, the claim that Chomsky made is that these various kinds of impossible languages, and they make up a bunch of different ones, should be as easy for the large language models to learn as human languages. And what they find in these simulation tests, where they force a poor little large language model to learn these, is that’s not really the case. They do struggle with the impossible ones. They do seem to learn them after some time to a certain extent.
MATT: Yeah.
HEDVIG: So, it’s different in ease of learning, but they are… Unless I misunderstood the paper wrong, they are learning them at some point.
MATT: Well, I mean, that’s a really odd result about neural networks, is that they can learn any function, the universal function approximators. So, it shouldn’t be no surprise that they can learn an arbitrary set of rules. As to how it does it, that’s a different question.
HEDVIG: Right.
DANIEL: So, I think I get the idea that Chomsky and company claim that even the most complicated patterns can be learned. Whereas this team, the Stanford team, is saying either no, they can’t, or it’s not as easy for them. Either way, what’s riding on this discussion? What difference does it make either way?
MATT: What is language? I think one interesting way to look at language is its structure. And if we’re going to talk about language, we want to know what makes language structure like it is. And if we think about what’s happening in this paper, we’re thinking about here are different ways of describing grammars, different ways of creating or structuring languages, some of which we know and some of which we don’t know. What are we doing when we’re using an LLM?
In the old days with neural networks, it used to be incredible if you could get them to learn anything at all, right? And it’d be like a huge result. It’s learned irregular pluralisation morphemes or whatever, and that’d be a huge result. And so that was the criterion back in the old days. “Look, you say that we need innate, language is like it is because that’s our innate linguistic faculty or whatever. But we got this neural network which doesn’t learn anything at all usually. But even that was able to do this,” was the old argument. Now, these networks are so good at learning that it’s kind of switched. And we want to think about what does it mean if something that’s really, really good at learning, has different abilities to learn different types of grammar?
HEDVIG: I’ve been thinking a lot recently about what you try and do when you speak language because what you’re trying to do is usually make someone else think a certain way or do a certain thing. You’re trying to coordinate action. And we talk about language sometimes as if that’s the main thing language does, which we know that it does other things too, but also as if that’s the only thing we have to do that, which it’s not the case. You can use all kinds of shared knowledge or gestures or inference to make people do what you want without speaking a single word.
MATT: Exactly.
HEDVIG: So, it’s not like languages has one job and it’s not like language is the only one doing that job even.
MATT: Yeah. I mean, I think possibly the reason that my academic career was somewhat abortive in the end, I hadn’t gone into industry is because I decided to be happy with the fact that nothing… there are no simple answers. I’m never going to be able to give a simple answer for what is language? Or, what is language for? Or, how did it evolve? Or, what are the parts of the brain which are involved in language? Because it’s just like this huge mix.
Think of all the things we use language for. Think of all the different types of language. Think of all the different ways of describing language, etc., etc. And I feel like the way that it just spans so many different phenomena and things we do and mechanisms and so on, that’s what is special about language, is that it just covers all of these different things. And I think that trying to get simple explanations for parts of that is great. But I think trying to get a simple explanation…
HEDVIG: For all of that.
MATT: …and then, tell other people that you’re not interested in anything else is not great. And that’s where I land on that.
DANIEL: Wow. We’re talking to Dr Matt Spike. Matt, this has been a fascinating conversation. Thank you so much for bringing what you know to the show.
MATT: Thank you for having me! For listening to me.
[LAUGHTER]
[INTERVIEW ENDS]
DANIEL: And now, it’s time for Related or Not. And our theme tune this time has been granted to us by Gordon. I think you’re going to love this. Let’s have a listen.
BEN: Three, two…
HEDVIG: I am clicking.
BEN: One, two, go.
[RELATED OR NOT THEME]
GORDON: [SINGING] Hey, get ready. It’s time to start the game.
Which of these words come from words which were the same?
Do they share a common ancestor, like the orange and kumquat?
Though it might be their similarity amount to diddly-squat.
Are they related or not?
HEDVIG: This is very cute.
GORDON: [SINGING] Can you see how they’re connected? Can you trace the dots?
You don’t need to be a linguist or even a polyglot.
Are they related or not?
It doesn’t matter if they sound the same or if they’re close semantically.
The question is whether they exist on the same branch of the etymological tree.
Are they related or not? (The aim of the game is in the name.)
Are they related or not? (Play it at home or on a train.)
Are they related or not? (Say it out loud or in your brain.)
Are they related or not?
HEDVIG: That was incredible. I love that.
BEN: Delightful.
DANIEL: It made me feel good because getting creative helps other people be creative too. I love how people are jumping in and sharing.
BEN: That reminds me of… Hedvig, you can help me out with this. The TikTok sound that was taken from a children’s TV show from a long time ago. [SINGING] Into the thick of it.
HEDVIG: [SINGING] Into the thick of it. Into the thick of it.
BEN: Can you remember the name of that show?
HEDVIG: No.
HEDVIG: Okay, so that was The Backyardigans. So, that reminds me…
HEDVIG: The Backyardigans. Oh, my god. Yes. Sorry, sorry. Backyardigans is amazing!
BEN: Yes. So, that reminds me of The Backyardigans, which I never caught when it was on, but all accounts, has a suite of music that goes harder than not just any other children’s TV show, but basically any TV show ever. It’s just like red hot… And honestly, that was giving such Backyardigans to me just then, which I don’t know if there’s a higher compliment for a piece of music. So, there you go.
HEDVIG: It’s very high. It’s very high.
DANIEL: Okay.
HEDVIG: If you have a child and they are between the ages of 2 and 5, Backyardigans is really fun.
DANIEL: Fantastic. Well, our first Related or Not comes from Elliot, who got in touch with us via SpeakPipe.
BEN: Oh, yay.
DANIEL: Let’s listen.
ELLIOT: Hello, Because Language. This is Elliot again. Since I’ve been listening and catching up on podcasts, I’ve spent all my free time thinking about the Related or Not game. So, I have three words, two of which are related, one of which is not. The first word is sale, S-A-L-E, as in to sell something. The second word is salary, S-A-L-A-R-Y. And the third word is salt, S-A-L-T. Which of the two are related?
HEDVIG: Mm-hmm.
DANIEL: Which of the two are related?
HEDVIG: I know something.
DANIEL: I know a thing too.
BEN: Oh, I have a feeling we might all know a thing on this one. If the thing that I’m thinking is the same thing that you’re thinking. Are you thinking what I’m thinking? Because I am thinking of a thing, are you thinking the same thing?
DANIEL: I am thinking what you’re thinking.
BEN: Okay.
HEDVIG: Okay. Should we all say it at the same time?
BEN: I’ll go first. No, no. I’ll go first.
HEDVIG: No, no, no. Let’s say it at the same time!
BEN: Okay.
DANIEL: Okay.
BEN: But how do we do it, to say it at the same time? Because it is two related…
DANIEL: We have to say the…
HEDVIG: We start with the noun phrase…
DANIEL: We have to say the one that’s not related, okay?
BEN: That’s it, yeah, yeah, yeah. That’s what we’ve got to do.
DANIEL: We’ll say 1, 2, 3 and we will say the odd one out.
HEDVIG: No, no, no. We all have to say what we think. We all have to say this myth we have in our heads.
BEN: But how do we say that…? Like, if we say the same word at the same time, then it’s actually going to sound…
HEDVIG: We try. We try. We try. Okay, you ready?
BEN: So, you want us to say the myth.
HEDVIG: Yes. Okay. Three, two, one.
DANIEL: Two, one.
[IN UNISON]
HEDVIG: Roman soldiers were paid in salt.
BEN: The salt is salary.
DANIEL: They used to pay Roman soldiers in salt.
[LAUGHTER]
BEN: Okay, there we go. It didn’t sound good, but we all were on the same page.
DANIEL: So, yes. The version of that was they used to pay Roman soldiers in salt. So, SALT and SALARY became related. But, Ben, what do you know?
BEN: No, it’s not what I know. I’ve heard that myth. What I’m wondering is, Elliot, that sneaky little bastard, does he know that this is a folk etymology, and he’s dropped this in here deliberately to fuck with us? And if he has, Elliot, you goddamn glorious bastard, I will doff so many hats to you, sir.
HEDVIG: Or as Elliot either doesn’t have that kind of machinations or double bluffing us, that’s the other problem.
BEN: [GASPS] [LAUGHS]
DANIEL: We’ll never know. We’ll never know.
HEDVIG: Either less Machiavellian that you think or more. [BEN LAUGHS]
DANIEL: Much more. Much more.
BEN: Okay, so what do we think…? I’m going to put my hand up and go, that I think SALE and SALARY are related and that the SALT and the SALARY thing is a folk etymology that Elliot has brought in. So, I choose less Machiavellian.
DANIEL: You’re just chucking it. You’re just chucking it all because of the folk etymology. Even though…
HEDVIG: You choose median Machiavellian. I choose extra Machiavellian.
BEN: So, you think it’s the same?
HEDVIG: No, I think that SALARY and SALT are related and that Elliot’s like double bluffing us, thinking, “They’re going to think the folk etymology isn’t true. But it is actually true.”
BEN: What would be really fun is if that is a folk etymology but they are related for a completely different reason? That’ll be really fun.
HEDVIG: That would also be fun. That would also be fun.
DANIEL: Well, the related pair are, as astute listeners will have realised, SALARY and SALT; SALE is the odd one out. And there is a folk belief, it seems to have arisen around the late 18th or early 19th century, that Roman soldiers were paid in salt. However, even though that’s not true, here’s the fact behind it. Salt was very useful in Roman times, and people did have to buy it because it was used to preserve food, meat or fish or olives.
BEN: Like a fundamental necessity in life.
DANIEL: So, it’s not that Roman soldiers were paid in salt. It was that some of the money really did have to go to salt and just as we talk about bread or dough…
HEDVIG: Daily bread. Yeah.
DANIEL: …in our culture. That’s slang seems to have arisen. Now, one thing that pushed this myth along is a mistranslation or a misinterpretation of a phrase in Pliny the Elder’s Natural History. Pliny says, and this is in translation, “Therefore, heaven knows, a civilised life is impossible without salt. And so necessary is this basic substance that its name is applied metaphorically, even to intense mental pleasures.” Which is why we have expressions about salts that are about the best of something.
BEN: Oh, salt of the earth and all that kind of stuff.
HEDVIG: Oh, salt of the earth.
DANIEL: Yeah. One spurious etymology that I found had SOLDIER being “sal dare”, to give salt to someone. That’s not where it comes from. SOLDIER is probably more likely to be from Latin solidus, the name of a Roman gold coin. It wasn’t made of cheap, flimsy metal. It was the thick stuff because you have to pay soldiers. So, SOLDIER comes from that. SELL and SALE are different. They both come from Norse sala. Big thanks, Elliot. That was a lot of fun.
BEN: Yeah, that was a good one.
HEDVIG: That was really good. Thank you.
DANIEL: Makes you feel good. The next one won’t. So, this one… [BEN LAUGHS]
HEDVIG: Please get in touch and say your level of Machiavellianism, if it was zero or extra.
DANIEL: Please do.
BEN: Yeah, a good point.
DANIEL: Please do. This one comes from Gordon, who granted us the lovely Related or Not jingle at the top of this segment. Got in touch via email, hello@becauselanguage.com. The two words are YELLOW and YOLK, as in the yolk of an egg.
HEDVIG: Oh.
BEN: YELLOW and YOLK.
DANIEL: Got to admit, that’s pretty yellow.
BEN: Not really.
DANIEL: [LAUGHS] It’s not?
BEN: No.
HEDVIG: It is.
BEN: No. Well, certainly not always like most people who keep their own chickens, which I assume is how most eggs would have been encountered back when these words were coming about, orange, really orange. Yellow is a very modern thing. [DANIEL LAUGHS]
HEDVIG: Right, Ben? That’s true. But how…
BEN: How old is orange? Mmm.
HEDVIG: How long do you think the yellow-orange split is? And before then, do you not think that orange was a part of yellow?
BEN: Good point, good point, good point, good point. Okay. Yeah.
HEDVIG: Because orange is basically just…
BEN: Redder yellow. Yeah.
HEDVIG: Yeah.
BEN: For most of human history, it was just like, “Oh, the redder yellow.”
HEDVIG: Yeah. We used to call it fire yellow in Swedish as brand gul, means like essentially orange.
BEN: That’s so cool. Oh, Swedish, stop it.
DANIEL: Let’s get back to the game. YELLOW and YOLK.
BEN: Yeah, yeah, yeah, yeah, YOLK and YELLOW.
DANIEL: Do you think they’re related or not? I said yes, because the white of the egg is albumen, a reference to the colour white, albus. So, I said, “Why not yellow?” Could be calling egg things their colour.
HEDVIG: I have some recollection that yolk is used in colour, pigmentation to make oil colours.
DANIEL: Okay.
HEDVIG: Is that true? But then, it’s probably just a binding agent.
DANIEL: Okay. The protein.
HEDVIG: I’m going to say that they are not related.
DANIEL: Okay.
BEN: I’m going to go with not as well. Far be it from me to side with Hedvig, but I feel like this could be one of these things where, yeah, just an unintentional similarity. I do, however, think they probably both come from the same root because they’ve got that quite guttural, I’m guessing, Germanic kind of space, like Norse, that sort of thing. Because they’ve got that really, like, [SOMEWHAT GUTTERAL] “yuolo” kind of thing going on, as opposed to the fruity romance languages with their [FLOWERY ROMANCE-LANGUAGE VOICE] sa se sa pe pa….
DANIEL: Ben, it sounds like what you’re describing is actually related, because…
BEN: No, no, no, as in being from the same root language isn’t the same as being related as a word.
DANIEL: Okay, okay. All right. Well, I say yes, you both say no. I’m right, and you got it wrong. They are related. Both words come from Old English. And we had geolu for yellow and geolca, which literally meant the yellow part. So, the yolk is just the yellow part, and YELLOW and YOLK are related. So, now knowing that YELLOW used to be geolu and there was a G to Y thing, can you guess — this is the bonus round — what other colour is related to yellow?
HEDVIG: Green.
DANIEL: Nope.
BEN: Grey?
DANIEL: Nope.
BEN: Oh, hang on. Should we be looking for Y words actually?
DANIEL: You should be looking for a G word.
BEN: Okay.
DANIEL: And it’s kind of yellow in a sense.
HEDVIG: It is not gray or green.
DANIEL: And it’s a metal.
HEDVIG: It’s a metal?
DANIEL: It’s a metal that is yellow.
HEDVIG: Copper. But that’s not… Gold, gold, gold, gold, gold, gold!
DANIEL: [LAUGHS] Hedvig gets it first.
[LAUGHTER]
It’s gold. They’re both from the Proto-Indo-European root, *ghel-, meaning to shine. Thanks to Mark Ellison, who gave me that one a while ago, and thanks to Gordon.
BEN: What was that about us being smart people? Honestly. [LAUGHTER] Way too wrong.
DANIEL: Well, I had you thinking about eggs, right? You were sidetracked.
BEN: Yeah.
HEDVIG: Yeah.
DANIEL: This one’s from Scott via email who says a sheep has a fleece and a group of sheep is a flock. So, FLEECE and FLOCK, related or not?
HEDVIG: I don’t know is the fleece only when you take the skin of, or is wool also fleece?
BEN: No, no, no. When you shear them.
HEDVIG: Right.
DANIEL: Yeah. It’s minus the skin explicitly, in my headcanon.
HEDVIG: Okay.
BEN: Yes, definitely. So, it’s the bit that you cut off without harming the sheep.
DANIEL: Otherwise, it’s hide. No, wait. Sheepskin. Sheepskin.
HEDVIG: Yeah, sheepskin. Like, what you have in front of a sweater.
BEN: I’m saying not.
DANIEL: Ben says no. Do you have a reason?
BEN: And I’m saying no for the same reason as I said no the previous time or kind of the same reason. Fleece with “ece” at the end and flock with an “ock” sound like they come from really different places linguistically. So, I’m going, no, the “fl” is… I don’t know, I feel like that’s a red herring.
DANIEL: Okay, I’ll go next. I thought the same thing as you, but I also realised that there are some forms that are related that do have that kind of thing. And the thing that I thought was causative DRINK and DRENCH, which used to be “to cause someone to drink”.
BEN: But still “ench” is still not… Like, “eece” just has that really Romantic kind of feel to it.
HEDVIG: And also, DRINK and DRENCH are more connected than semantically than FLEECE and FLOCK. Or you mean like part. Oh, you mean like part and whole or something.
DANIEL: Something like that. I don’t know.
BEN: Okay, weigh in, Hedvig, what do you reckon?
HEDVIG: All right. Okay, so I do know that FLOCK is “flock” in Swedish and other things flock that aren’t sheep like birds…
DANIEL: Birds could be flocks.
HEDVIG: So, that’s somewhat relevant.
DANIEL: She’s having a tough one.
BEN: She’s really ruminating on this one.
HEDVIG: I think no.
DANIEL: Okay, so you both said no again, and I said yes again. And this time, I was wrong. You both got it. They’re not related. FLEECE comes from West Germanic Old English fleos, which probably comes from Proto-Indo European pleos, to pluck. That’s also where we get pluma in Latin, which is a feather.
BEN: Right, right.
DANIEL: Whereas FLOCK just comes from the Old Norse flokkr. And it could be for birds as well as for sheep.
BEN: It makes sense. As Hedvig was describing that, I was like, if you’ve ever seen a mob of sheep on a hillside getting rounded up or whatever, they really do look just like a flock of birds in the sky. Like, they move in… They have that really similar visual look to them. So, you can absolutely see how people would be like: flock… flock.
HEDVIG: They do the thing that the fish do as well, where they like…
BEN: Yeah, exactly.
DANIEL: Yep. Well, thank you, Scott, for that question. And as always, thank you everyone for sending these. We’re working through them and we’re having lots of fun. If you’ve got a Related or Not idea or a jingle you’d like to give us — thank you, Gordon — why don’t you send that to us at hello@becauselanguage.com or any of the other ways that you can get in touch with us.
[MUSIC]
[INTERVIEW BEGINS]
I’m here with Dr Julie Sedivy, author of many books including Memory Speaks and most recently, Linguaphile: A Life of Language Love. Hi, Julie. Thanks for coming on the show.
JULIE SEDIVY: Oh, it’s such a delight. Thanks for having me.
DANIEL: This book is really, really different. I don’t know what I was expecting, but… Does everyone say that, by the way, “Wow, this book is so different”?
JULIE: [LAUGHS] I think this book is different. So yeah, I’m not surprised to hear you say that.
DANIEL: Well, it’s different because it’s so personal. I’m just reading the genre here, and it’s listed under: Language, Memoir. I don’t think I’ve ever seen those two together. What made you decide to approach this discussion of language in such a personal way?
JULIE: Well, it is a fusion. I really struggle actually to describe this book to people when they ask me about it. So, I’d be really curious to know how you would describe it.
DANIEL: I describe it as poetry, to be honest.
JULIE: Oh, wow. Wow. Okay. I love that.
DANIEL: Okay. I wasn’t going to say this, but I wrote this down in my notes. It’s like, “This is like a linguistics textbook.” And then, I crossed out textbook. “This is like a linguistics book, but in the form of poetry.”
JULIE: I love that. I love that. So, I wrote this book for a very specific reader, actually. I wrote this book for a student who showed up in one of my Intro to Linguistics classes, stayed for the first lecture, and then came up to me at the end of the class and said that she was not going to be continuing in the class. And I mention this story in the book. And by way of explanation, she told me that she was a poet and therefore she was not interested in dissecting language in the way that it seemed that we were going to do in the class. And I was utterly dumbfounded because my experience of studying linguistics and psycholinguistics was that it just deepened the beauty and mystery of my experience of language.
So, she has been rattling around in my brain for decades, and she’s the person I had in mind when I wanted to write this particular book. I wanted to express something about my relationship to language, and my relationship to language as augmented by a scientific knowledge of it as well. But I knew that to do that in an informational, more traditional, lingcomm manner probably wasn’t going to win her over and wasn’t also going to express really authentically my own relationship to language. So, I knew that the book was going to be different.
Also, I’m really drawn to the genre of memoir. Not the celebrity memoir or the famous person who’s had a fabulous life and you want to know all the dirty details, but the kind of memoir where someone just sifts through the ingredients of their life and tries to make sense of life by looking at their life. And that was kind of what I was hoping to achieve. And it just so happens that many of the ingredients of my own life happen to include lots of signs of language in it.
DANIEL: It’s funny that student had that reaction, because usually what happens is, they come in knowing nothing about linguistics… By the way, is teaching first years not the best thing in the world?
JULIE: It can be. Yeah, absolutely. Absolutely. Yeah.
DANIEL: So, they walk in and they go, “This is amazing. I’m changing my entire life now,” which is your experience. It’s my experience. I think it’s for a lot of us.
JULIE: Hey, yeah, absolutely, yeah.
DANIEL: Do you find that in treating your own life and then also language, did you find that you were split between the scientific and the personal, or did you find that they just meshed very easily?
JULIE: In the past, I have felt split. I think it took me 10 years to become a good enough writer to write this book. And I know that I remember when I first started writing about language science in a way that was for a broader audience, I felt like I would write Frankenstein pieces where there were the lyrical parts, and then there was the part where I described the experiment, and I would just lapse into the formulaic scientific language and really struggle with making a coherent whole. I think, for the most part, I have integrated those two parts of myself in style of language. So, this felt more organic in a way that was really, really satisfying to me.
DANIEL: So, it was kind of like bringing your life together.
JULIE: Yeah.
DANIEL: Parts of your life.
JULIE: Yeah. To me, this book feels like an integration of different parts of my life. I started out assuming that I would be a novelist because I loved literature, I loved language, I loved reading and writing. And then, yeah, that fateful linguistics class that just kind of knocked me off course and set me onto a scientific path. So, this really represents a coming back to something that I had always felt was an important part of my life but had been underdeveloped for a couple of decades.
DANIEL: When I was reading Linguaphile, I found myself trying to get back to the scientific. I was like, “I’m enjoying this discussion of life. Now, there’s a linguistic bit. Oh, good. I’m comfortable here. I can retreat into it.” And I was trying to figure you out how, when you take a quiz online, it’s like, oh, what ’90s rock star are you? You look at the questions and you’re trying to figure out, oh, what is this question going to point us to? I was looking at the bits and saying, “Okay, what is Julie’s work pointing to?” You started out in there’s a lot of psycholinguistics in your past. And I thought about my own linguistic life. And I’ve gone from computational linguistics to sociolinguistics to recreational linguistics, and I’ve come around to cognitive linguistics. Have you moved around as well?
JULIE: In my research life, no, not really. I was really solidly a psycholinguist in my research. I wrote a 600-page textbook on psycholinguistics.
DANIEL: Yes, you did.
JULIE: So, that, I think, forms the borders of my professional life. But I really do enjoy reading outside, particularly in sociolinguistics. I think that’s just a fascinating field of study.
DANIEL: Yeah. Okay, so now I’m going to ask you that question. What is language?
JULIE: [LAUGHS] What is language? Well, you sent me a note earlier. You said, “Okay, I know how you’re going to answer.” And you provided a quote that actually came from Helen Keller’s memoir. There’s a memoir that really reflects a lot of reflection and introspection as well. So, let me remember what she said. Actually, she defined love as the invisible threads between our spirits and the spirits of others. And when I read that, it was in the context of reading her talking about her relationship to her teacher, Ann Sullivan, and just like the momentous opening that language offered her. And when she learned about love, this was the definition that she had.
And to me, it just felt like this is also language. This is what allows us to share minds with each other in the level of detail that we can. It’s hard to imagine that we could possibly be so intimate with each other if we didn’t have language.
DANIEL: There’s a real connection between language and love. And when we’re talking to people that we love, that language is going to be different from people that we’re not as connected to. Somebody asked me once, “What is love?” And I said, “As a linguist, my answer would be alignment.”
JULIE: Mm. [LAUGHS] Yeah.
DANIEL: When you start a conversation with somebody, you say, “Oh, let’s just have a brief bit about what this conversation is going to be about,” and then you’re aligned. But I think alignment is like, when you love a partner, your lives are aligned. When you love a child, your life is aligned with their life. When you love a friend, you have a history, a shared history, and you’re pointed in the same direction. So, I really do think of language and love to have those commonalities.
JULIE: Yeah. And of course, alignment is a very powerful concept in psycholinguistics as well. And you see it in the very, very early stages of language learning. For instance, there’s this fascinating research that shows that even just listening to the sounds of a foreign language, that children learn the distributional patterns of that more quickly if they’re hearing it from a person that they’re interacting with, than they hear it over a loudspeaker or even see it on a video screen. So, there’s something about that human connection, that alignment of attentional focus, that sense of, I don’t know, something we don’t know exactly what it is, but it seems to be a powerful ingredient in language learning and in language use, in language form. I think sociolinguistics is also to a great extent is about alignment.
DANIEL: Our last episode was a bonus episode about the one thing that explains the most about language. And my answer — spoiler alert — was language looks the way it does because of human cognitive limitations. That to get over the limitations of our memory and our attentional focus and our ability to understand, we have to do certain hacks. And language, you could profitably explore those hacks. But what would you say is one really important thing that explains a lot about language?
JULIE: Yeah. Okay, can I have two?
DANIEL: You get two. I’ll give you two. [JULIE LAUGHS] They might be the same thing. They might be part of the same thing.
JULIE: They certainly interact with each other, yes. The first would be what we’ve just talked about, which is, I think, alignment and the drive to align minds with each other. So, that, I think, is responsible for the powerful… just the push that we have to communicate with each other. The way that we can manage our attention jointly with another person, that’s an absolutely essential ingredient to learning language, to using language, to navigating our way through a complex discourse while maintaining shared attention on the same elements, all kinds of things.
And then the second, I would agree with you, has to do with our cognitive limitations, and specifically the limitations that arise as a function of language being something that takes place in sequential time. We have all of these complicated elements and relationships that we want to express in a sentence. We’ve got lots of nouns and verbs and subjects and objects and prepositional phrases and sometimes clauses within clauses. Maybe if we had many appendages with elaborate hand shapes, we could express all of this simultaneously. But given our bodies and the tools at our disposal, if we use voices, we’re going to be doing that in linear time. And then, that brings in all of the limitations of memory that you mentioned, brings in the limitations on inference and prediction that really shape the form that language takes. So, I would say those two things together really give language the form that it has.
DANIEL: Yeah, that’s an amazing answer. I would fully agree. You spent some time in the book talking about crevasses in language, what kind of crevasses there are. How are they like language and how are they dangerous? Is language dangerous?
JULIE: [LAUGHS] Yes, it can be, as we all know. [LAUGHTER] Yeah. So, the crevasses refer specifically to the gaps that are left by language. We never fully put into language everything that we mean to say, and here certainly recover a lot more from what we say than has been said. I start that chapter off with anecdote in which I remember asking my mother at the age of 12, “Mom, do you think I’m pretty?” And her answer was, “It’s more important to be smart than to be pretty.”
DANIEL: Oh, no.
[LAUGHTER]
JULIE: So, there is the crevasse into which I fell. I think it’s easy to imagine what might have been going through my mind when I received that answer. So, obviously, this is a very useful function for language. It allows language to be superefficient that you can communicate to me something that you intend, and I can often recover that intended meaning very accurately.
But it also leaves all kinds of room for misinterpretations. How many of us have not lain awake at night ruminating over somebody’s intention and turning over their words in our mind and wondering what did mean by that? Or misunderstandings that have been difficult, or workplace frictions, because maybe we belong to different cultures, and some of us come from cultures where we expect problems to be expressed very directly, and others of us come from cultures where these things are not aired in a direct manner, but are just kind of suggested in very subtle ways. So, there are all kinds of dangers lurking in these gaps between languages that I think are really interesting to explore and acknowledge.
DANIEL: And you mention also that when people are climbing around in snowy places where crevasses might lurk, they depend on each other to keep each other safe.
JULIE: Yes, yes.
DANIEL: When I say something very elliptically or when make an implication and I got to make implications because otherwise it would take too long, I’ve got to make my words do more in less time so that you’re not bored as my listener. But I’ve got to trust you that if I put it out there, you’ll be able to draw the correct implication, that you’re competent as a speaker.
JULIE: Exactly. And this is where the alignment part of language again comes in, is that we have to have this constant attunement and we have ways of signaling that often in an interpersonal interaction. Especially when we’re face to face with each other, that gives us a cue, as to how well the hearer is recovering the intended message or not. So, yeah, the controlling metaphor of the crevasse really refers to the dangers, but yes, also the safety mechanisms that we have of roping up to each other when traveling over this treacherous terrain and making sure that we don’t fall in.
DANIEL: Got to keep each other safe.
JULIE: Got to keep each other safe, yeah.
DANIEL: Reminds me of when I talked to Dr Nick Enfield in an episode. It was episode 73, Consequences of Language. And he mentioned an essay by Margaret Gilbert called Walking Together. It’s about how when we are walking with someone, we accept that there’s a responsibility to each other. I’m not going to walk too fast. I’m holding my child’s hand, and I don’t want to walk too fast, or I’m walking with my partner. And so, do I decide to shift my focus to left, right to match her left, right. If we have our arm around each other or something, we have to walk differently.
And those responsibilities just don’t exist if you happen to be in a crowded place walking next to somebody. There’s no expectation that we’re going to walk together or that we’re going to make allowances for each other or that we’re going to match our stride or anything like that. But when you walk with somebody, it does mean to accept a certain kind of responsibility. And Nick made this point that we have responsibilities to each other in language as well. So, that’s what that part made me think of.
JULIE: Yeah, yeah, very much so. Very much so.
DANIEL: I want to ask about the metaphors you use for language, like a crevasse or a painting of an autopsy [JULIE LAUGHS] or a fruit drop. There are a lot of very delicious metaphors, and I wondered about how you came to be skillful in choosing these things. Did you find that was a lot of work? Or, why did you use metaphors in such a way?
JULIE: Well, because I think a language is more powerful if it’s very specific. And here, I think I wanted to exploit some of the beauty and the power that comes from literary styles of language that are often in some tension with scientific styles of language. So, in some ways, I think that this book has some of the characteristics of maybe an impressionistic painting. It depicts a subject, but it’s not meant to be an anatomically accurate depiction, a clinical depiction. It is an impressionistic creation that is meant to create a certain mood, to give pleasure was one of my main goals for this book, was that the reader experience a sense of pleasure while reading it.
So, I really drew on some of those techniques, and that was part of the process of developing the skills I needed to get to be able to write this book. So, some of that process involves writing lots and lots of poetry, which I am not interested in publishing because I don’t think that’s where my contribution as a writer is. But it allowed me to approach writing using a different process, and with a greater comfort level for some of the unconscious associations that crop up when you’re just playing with a poem.
So, this book did not have an outline. It really was just something that I allowed spontaneously to bubble up and then carefully chose the things that seemed to be working and moved language around and really carefully deliberated over very specific words and the rhythm of phrases. So, I loved when you described this book as like a textbook in psycholinguistics, but in poetry, because I felt that was the skill set that I was really drawing upon when working on the book.
DANIEL: Well, I think that came through, and it also really worked. I think that you’ve managed to arrange it into the sections that you did, the beginning of life, the middle, and then loss, not just individual loss of a person, but language loss. So, language arises, language exists, and then it fades. An individual language or an individual word, or the number of listens that this episode of the podcast will get.
JULIE: Oh, how sad!
DANIEL: It is. And yet, we struggle for permanency, don’t we? All of our words fade, but we try to capture them. We try to grab them.
JULIE: Yeah, yeah. And yet, if language is nothing, if not evanescent, and to some extent, this book is a coming to grips with the evanescence of everything.
DANIEL: I don’t like that.
JULIE: Mm.
[LONG PAUSE]
DANIEL: There’s a quote from the book that I think really puts this well. I’ll just read it, “In this back and forth, pressed between the walls of past and future, as they close in on each other, language seems a fragile thing in constant danger of being crushed.” How is that the case?
JULIE: So, that came from a chapter called The Rectilinear Movement of Time that really talks about the implications of language being instantiated in sequential time and all of the limitations that we alluded to earlier having to do with memory and our cognitive powers. As anyone who’s taken a psycholinguistics class knows, if you’re listening to language, you’re immediately trying to gather up the elements of language, assign some meaning to it. And because that’s going to be indeterminate at every slice of time, you don’t know how the sentence is going to continue. You’re trying to project yourself into the future and make guesses as to what trajectory the sentence is going to take. So, you have to be doing that prediction before what has already been uttered fades from your memory so that you can no longer accurately predict the future. So, it’s this reality that we are looking at every moment of our lives through this tiny, tiny pinhole of the present, but trying to do something that essentially extends far beyond the scope of that present. And language is just a beautiful example of that.
DANIEL: And you mentioned that from the listener perspective, but I think it’s also true from the speaker. And I know that we don’t want to exclude signed language and written language, but I’ll just focus on speaking and signing for a second. The listener has a wide range of possible meanings, and they quickly learn to exclude what isn’t relevant, syllable by syllable, as it goes on.
JULIE: That’s right. That’s right.
DANIEL: But the speaker or signer is also doing the same thing. I have just launched into a sentence and I have kind of a plan for it, but I’m not sure exactly how it’s going to end or what I’m going to do with it when it’s done. So really, we are coming together and sort of scrabbling together these pieces.
JULIE: Yeah, exactly. And you may have forgotten how it started by the time you get to the point where you figured out what it is that you want to say. Yeah, absolutely, I think the pressures on the speaker, if anything, are greater than the pressures on the hearer.
DANIEL: And that sounds right in line with what you must have discovered in your psycholinguistic research.
JULIE: Mm-hmm. Yes.
DANIEL: Since we’re talking about ambiguity, do you think that I could get you to read a bit for me?
JULIE: Sure, sure. I’d love to. So, this is in a chapter called Resolving Ambiguities, which happens to have… the study of ambiguity has taken up a very large space in my research life, actually. So, it’s something that I know about intimately. And as anyone who takes a psycholinguistics class discovers, language is awash in ambiguity. And because we’re constantly trying to structure meaning on the fly, we’re often faced with huge indeterminacy, even syllable by syllable. The syllable cap, for example, could continue as captain, capture, captivate, cappuccino, captain, and so on. So, we’re trying to make these moment-by-moment decisions related to meaning in the absence of good evidence.
And one of the things that struck me is how, despite the fact that ambiguity is so present in language, outside of the domain of psycholinguistics, most of the time we think of ambiguity as an undesirable thing, as a stressful thing. And that’s what really captivated me. And I wanted to explore that in the chapter. So, I structured the chapter around the experience of seeing two viewings of a film at different points in my first marriage and having the feeling that the second time I saw it, I had a completely different interpretation from the first. And that was the entry point into the chapter. So, towards the end, this is the segment that I’m going to read.
“It dazzles me still that we humans ever manage to understand each other at all. It is a daily miracle, the way we can live amid the buzz of a thousand possible meanings and still manage well enough to settle on meanings that are mutually agreeable. It reveals an astonishing agility. With minor variations, we are all equipped with the mind whose aperture can widen to semantic possibility and then quickly contract to admit but one meaning. Unknowingly, we perform these acts of opening, selecting, suppressing countless times a day, like the rhythmic openings and closings of gills on fish, unbothered by the ambiguities in which we swim.
“How is it then that we can be so undone by certain ambiguities, precisely those that perturb our sense of who we are in relation to others? We are undone by them whether our temperament is to wander about in open semantic fields or take cover in confined spaces like my former husband. With these dangerously personal meanings, we struggle to find our rhythm. We dilate too wide and drown in unresolvable contradictions. Or we shut down too soon, deluding ourselves with certainty.”
DANIEL: It’s a dance, isn’t it, between the specific and the general, between all the possibilities and not enough possibilities.
JULIE: Mm-hmm. Yeah, indeterminacy is so baked into language, and we just navigate that space constantly so that, yeah, I think is just a fascinating property of language that struck me as having all kinds of metaphorical possibilities for talking about life.
DANIEL: I think so too. I think I left something out of my earlier discussion about language as a series of hacks to overcome cognitive limitations. Reality is huge and it’s just too much for us. And we have to very quickly decide what to ignore. And we’re super good at deciding what’s worth ignoring. Not everything is worth the attention, and we can’t focus on everything. So, we have to decide, and we do it. We’re so good at this. There were lots and lots of little revelations like that in Linguaphile. Lots of parts where I recognised a lot about language, yes, but I also recognised some great and some painful things about myself and my own life. I think that’s a journey that awaits the reader, anyone who decides to pick up Linguaphile.
JULIE: Lovely.
DANIEL: The book is Linguaphile. It’s available now. Is it available now?
JULIE: It is available as of October 15th.
DANIEL: By the time you hear this, it’ll be available from Farrar, Straus and Giroux. And I’m talking to the author, Dr Julie Sedivy. Julie, how can people find you and find out what you’re doing?
JULIE: They can find me on Twitter X. They can find my web page and send me a note. They can stalk me in Calgary and find me on the hiking trails.
DANIEL: We don’t recommend that one.
JULIE: Those are the best options.
[LAUGHTER]
If you can find me on the hiking trail, I don’t know, I think your efforts might be rewarded with, I don’t know, something, maybe a conversation about language.
DANIEL: You can walk together. You can be aligned.
JULIE: We can be aligned.
DANIEL: Julie, thank you so much for hanging out with me and aligning a bit of our day today. I feel like I learned a lot from your book, and I really enjoyed it.
JULIE: Thank you so much for having me, Daniel. It was a great pleasure.
[INTERVIEW ENDS]
[MUSIC]
DANIEL: And now it’s Words of the Week.
BEN: Woohoo.
DANIEL: The first one comes to us from Andy from Logophilius via email. This was also suggested by Ann on our Discord. So, let’s hear what Andy says. “Hey, Because Lang team, I’m starting to see ‘sanewashing’ in the news.”
BEN: Oh, I can guess what this is.
[LAUGHTER]
DANIEL: “Which I define as news organisations rationalising the irrational statements of a certain orange menace, which seems like a pertinent Word of the Week. I don’t recall you talking about washing as a productive suffix before, but that could be a good discussion as well.” Was that what you were thinking, Ben?
BEN: Yeah, 100%. Just the idea that there is… I was going to use the word “cottage industry”, but actually it’s like an industry-industry of various media personalities who are attempting to render normal… just really what can only be described as truly bizarre behaviour and points. And obviously, we’re all like super anti-Trump on this podcast, but it’s actually gotten to the point now where it feels like it’s graduated even beyond that. I’m literally sort of the more things that are happening, I’m now at a point of like, “Can someone hospitalise this man?” I have genuine concerns over this human’s mental state because it’s a level of dysfunction that is approaching pretty wild levels. And then, people being like, “No, no, no. But that is a person who should be president.” And it’s like, “How?”
DANIEL: How?
BEN: [LAUGHS] “How could you possibly conclude that?”
DANIEL: It’s just dangerous. The statements of an individual can cause harm to loads and loads of people.
BEN: Yeah, it’s wild. Yeah. So that is exactly what I thought sanewashing was.
HEDVIG: I think his improvised, impromptu crazy ramblings appeal to a lot of people because they make him seem natural and approachable. Like, “My uncle also rambles a bit like that and I can kind of…” There’s this thing he does where he says a lot of things and he lets people fill in the blanks. And people really like that because then they can fill in blanks with whatever they want. So, I’m guessing things like they’re eating the cats and dogs are people are saying. “Well, actually, what he meant is…” and it’s like a metaphor for their something. I don’t know.
I was in Greece on a honeymoon when the American presidential debate was airing, and we got up to watch it. And the day after, I was at a very nice massage appointment. And she’s like, “Oh, you’re very tired.” Yeah. We went up and watched the presidential debate, and she was just like, “Why…?”
[LAUGHTER]
BEN: Why? [LAUGHS]
HEDVIG: We had to be, like…
BEN: Why would you subject yourself?
HEDVIG: “Uh, it’s interesting.” And she was like, “Uh-huh. What? What are you people?”
BEN: It’s essentially like, at the end of the day, it’s gotten to the point now where you can in good faith say to a person, essentially, what we’re watching is the highest stakes reality TV show competition that has ever happened.
DANIEL: We’ve seen lots of examples of the media using normal language to describe dangerous words and dangerous behaviour. On Twitter, @MarkJacob16, is on the case link in the show notes for this episode. becauselanguage.com. When Trump said that “Immigrants were animals, who would come into your house and kill you,” Bloomberg responded this way in the headline, “Donald Trump sharpened his criticism on border security in a swing state visit, playing up a political vulnerability for Kamala Harris.” So, this is bad for her somehow.
BEN: Can I throw in an example? So, I think we could find basically infinity examples of this being ridiculous, but can I throw an example of this being actually really genuinely pernicious and sly and pretty fucked up?
DANIEL: Yes, please.
BEN: It’s not to do with Donald Trump, I’m sure a lot of our listeners, but maybe you guys, maybe not. Recently famous liberal African American writer, Ta-Nehisi Coates, was being interviewed on CBS Morning News about a new book that he is releasing in like a week that is incredibly critical of the apartheid state that is Israel. And he was brought on to be essentially asked questions about his book. But it was by any rational person’s categorisation, a hit job. Ta-Nehisi Coates, the author, was just there to be accused literally and explicitly of holding terrorist beliefs, that he ascribes to a philosophy that… I believe the direct quote is “would belong in the backpack of a suicide bomber.” All of that is not what I’m talking about to be clear.
The version of this sanewashing thing that I think we all need to be incredibly wary of or at least looking out for, is the fact that the host who was interviewing Ta-Nehisi Coates was white. And so, the people who produced the show understood that there would be kind of not great optics on a white host going really hard on their guest and being pretty unforgiving in their line of questioning and that sort of thing. So, what they did was they flanked the host with two other Black men cohosts who had no job and had nothing to say and nothing to contribute, but simply provided this air of legitimacy so that if anyone said, “This is a pretty fucked up and racist thing to do to Ta-Nehisi Coates,” they were like, “Well, if you look at the screen, there’s three Black people and one white person. So, I don’t know what you’re talking about.” That is the really subtle sort of like… I know we don’t love Noam Chomsky linguistically, but like, politically, that’s straight out of Manufacturing Consent.
But that’s the stuff, that’s the kind of sanewashing type thing that we need to be really careful of. What are the little tiny things that are being done to give the impression of legitimacy, when in fact there really isn’t any legitimacy to a particular argument?
HEDVIG: I think that’s very interesting. I’m not sure, is that sanewashing? Is that what the word means?
BEN: Well, I don’t know if about sanewashing, but certainly affording what is actually a pretty extreme position the impression that it is no longer an extreme position. It’s sanewashing-adjacent, I reckon, or perhaps I think the point I was trying to make is sanewashing, when applied to someone like Donald Trump, is always going to look ridiculous and overt as people bend themselves over in knots and backwards to try and justify insanity. But there’s a bunch of stuff that is actually still right squarely within acceptable norms that we still should be fighting against and talking about that is much easier to sane wash because all you need to do is just like nibble around the edges a little bit and then things seem like, “Yeah, totally, man, 100%. I totally get where you come from,” that sort of thing.
DANIEL: There’s one thing that’s been bugging me about this term. I get what it means, but the way we talk about mental illness just sucks sometimes. I mean, Donald Trump says horrible things that harm people because he’s a horrible person who wants to harm people, not because he is mentally ill or insane in any way. And I know we know this, and we’ve talked about this before, but I think it’s a good reminder. People who work in the mental health space have pointed out that in their view, using ableist language around mental health can delay treatment for people because they say: “Insanity is stupidity. I’m not stupid and therefore I don’t need help.” So, I don’t think that we’re bad people or that you’re a bad person if you use the term SANEWASHING. Maybe though, let’s just think about the impact that our words have and the way that our words reflect beliefs about mental illness. And I think it’s a good idea to separate being a horrible person from being in a bad way.
BEN: Insane.
HEDVIG: But do you think that the way forward there is to get people to think differently about words like insanity or to be more productive in being more specific and say, this person has anxiety, this person has psychotic episodes, this person is manic depressive and being more specific and maybe… I mean, it feels bad, but like maybe giving up on INSANE as a term to use in any meaningful way.
DANIEL: I would love that. I would love to cut off the term INSANE and SANE from our discussion of mental illness. I don’t know if that’s going to happen, but I would love that.
BEN: I mean, we must be getting close, right? Whatever the treadmill is called…
DANIEL: The euphemism cycle.
BEN: …the semantic treadmill or whatever, like MORON and DOLT and all that sort of stuff, which were clinical terms have not been used for decades and decades and decades. So, we can’t be far away. But then the question becomes, how do we prevent the next one just being slotted into that position? And I think that’s what you were talking about, Hedvig. It was like, we just need to not be lazy and be like, “Oh, we need a catch-all term for all the kinds of psychology. We just find people yucky.”
HEDVIG: Not even that similar. [LAUGHTER] Like, illness comes in many different flavors. They’re very different. And none of them are just all like the same kind of irrationality by thought, by any means.
DANIEL: The euphemism cycle seems to stop when attitudes improve about the people being discussed. So maybe, we all can do some work there. Thanks, Andy, for that term.
BEN: What’s next?
DANIEL: This next one suggested by Pontus — affectionately known as Moon Moon — on Facebook. It’s W and L. Pontus says, “Did you ever talk about the newish adjectives W and L? For example, Daniel has W rizz. I think it’s the first one letter adjective in English.” Help me out, people.
HEDVIG: Wait, so, so…
DANIEL: Come on, high school teacher.
BEN: Okay. So, as always is the case with “cool language,” this has just been ferociously stolen straight out of the African American community. It literally just means win and loss. My understanding, and this is not a good one, is that it probably comes from sports that tend to be dominated by African American players. So, I think it might have come by way of basketball, like having a W, putting Ls on the board, all that kind of stuff.
HEDVIG: I don’t understand something here because how does it work in Daniel has W rizz. Daniel has win rizz?
DANIEL: Winning.
BEN: Daniel’s winning because he has rizz or Daniel’s rizz causes him to win and that sort of stuff.
HEDVIG: Huh, okay.
DANIEL: Yeah, and I am seeing L’s and W’s used everywhere. Like, take the L or posting your Ws or things like that. So, yeah, “W rizz and L rizz” to describe a person’s winning or losing abilities at attracting those special people.
HEDVIG: I have a comment on… Pontus wrote, “I think this is the first one-letter adjective in English,” I have heard people use V. Like, that is V interesting instead of very.
DANIEL: Oh, yeah.
BEN: Oh, yeah.
DANIEL: Bridget Jones’s Diary.
BEN: And if we go back a long way, V for victory as well.
DANIEL: Yep.
BEN: I think it was used that way back in the day.
DANIEL: V interesting. At least Bridget Jones’s Diary was the first time I ever saw V being used.
HEDVIG: Yeah, an adjective, like very. Yeah, yeah. That is V interesting.
BEN: Can I throw out there? So, for those who are listening to this and they’re like, “Oh, they’re young people and their language, it’s always changing.” Actually, the sense that I’m getting is these are falling out of use already.
DANIEL: Oh, no. [LAUGHS]
BEN: And aura, A-U-R-A is coming in to take its place.
HEDVIG: No, that’s…
DANIEL: Oh, how so?
BEN: You don’t think so, Hedvig?
BEN: I think AURA is a bit old as well. No?
DANIEL: Okay, well, you’ve got to fill me in, because this isn’t a usage I’m familiar with.
BEN: Basically, AURA is saying something lacks aura or it has lost aura or something like that is basically communicating that a person has suffered some sort of very embarrassing loss or embarrassment and the reverse. So, if you gain aura or gain aura points, it means you’re very capable or you’re flexing or something like that.
DANIEL: Fantastic.
BEN: I’m seeing W and L used a bit less. And I’m seeing AURA used a bit more.
HEDVIG: I have to say, we need to remind everyone here that we’re talking about youth language, and we’re talking about things that have a very short lifespan. So, as soon as it gets on mine and Ben’s radar, it’s basically probably on the way out. By the time, Daniel hears of it and brings it up, it’s very gone. [LAUGHTER] But also, remember young people who speak English, this is probably less than 1% of their… If you take all the words they use in a day, how many of the words are these?
BEN: Yeah, fractions of a percent.
HEDVIG: Most of the time, they say, “I don’t like potato salad,” and other things that everyone says. Like, we’re talking about very few…
BEN: Who says that, Hedvig? Come on.
[LAUGHTER]
HEDVIG: I’m just remind…
BEN: No one. Literally, no one’s ever said that.
HEDVIG: I am just reminding us all that the youth are also speaking very similar to us.
BEN: Human beings who are just trying to get through life and eat potato salad or not.
HEDVIG: Yes.
DANIEL: Well, I’d just like it to be pointed out that, yes, I may be slow on the uptake for language, but in example sentences, at least, I do have W rizz. Thanks, Pontus. Next one, BLUE ZONE. Anyone know what a blue zone is?
HEDVIG: The date…?
BEN: No. Is this where we take the Smurfs when we institute pogroms?
HEDVIG: I know! I know. I know, I know.
DANIEL: No, it’s not. But Hedvig knows. Oh, geez, thanks, Ben. Glad you could raise that spectre.
HEDVIG: It is Nobel Prize Month, but it is also Ig Nobel Prize time.
DANIEL: Yes. It is.
HEDVIG: So, this is the time of the year when various committees announced who gets a Nobel Prize in literature and peace and all the other things. And also, Ig Noble, which is a prize for science that makes you laugh and then makes you think. So, it can’t just be things that are funny. It has to have some sort of extra thing too. And research by this guy called Saul Justin Newman showed that in a lot of places where people say that people live for a very long time, where there are a lot of super centennials. So, people who are above a hundred are also places where a lot of birth records are hard to find and where there’s a lot of pension fraud.
[LAUGHTER]
DANIEL: So, they’re lying, basically. They’re lying or they’re mistaken.
HEDVIG: They were lying. Yeah.
BEN: Oh, so Okinawa is just a bunch of old people just ripping the government off. I love it.
HEDVIG: Places that have had bad infrastructure or have been severely bombed, it is hard to find archives of who was born when.
BEN: Okay, okay, I like it.
DANIEL: So, the blue zones. This has been a term popularised by researchers, Michel Poulain and Dan Buettner. They’ve identified some places where people seem to be not where the people generally have a long life. In fact, this is very negatively correlated, but where there seem to be a high proportion of supercentenarians. Okinawa, Japan, you pointed out Okinawa. Good job, Ben.
HEDVIG: Turkey.
DANIEL: Ikaria, Greece. Costa Rica.
HEDVIG: I thought Turkey, but maybe that’s wrong.
DANIEL: I think there’s a couple there. So, what happens is it’s highly correlated with low lifespan, with poor record keeping. Supercentenarians tend to have birthdays that are on the first of the month or some multiple of 5, and where pension fraud is rife. So, your dad dies at 90 and you’re 70, you step in, then suddenly you’re 90, but everybody thinks you’re 110. And then the news starts coming around and suddenly you’re the oldest person in the world and you’re like, “Doot-de-doo. I never thought it was going to get up to this level. It is really out of control.”
HEDVIG: Yeah. Please don’t pay any attention to me!
BEN: Do we not have a way to carbon date human beings? Can we not take a bone biopsy?
HEDVIG: We can.
BEN: Is that not possible?
HEDVIG: Teeth are pretty good.
DANIEL: That is a very good thing to do to an old person. Yes, I would love to do that.
BEN: Oh, fine.
DANIEL: [LAUGHS] So, this is research from Saul Justin Newman where he points out that, yes, the research on supercentenarians is just not very good. It’s a total mess. It is a new thing that I’ve become skeptical about is superannuated humans, over 110 especially.
Okay, and finally, last one from Gordon. ATE THAT. Ate that. Gordon says — Gordon, it’s you again — “TO ATE is popping up all over my Twitter feed at the moment. I don’t really know what it means, but if you search Twitter for ‘ate that,’ you’ll get plenty of examples.” So, I did. Let’s see: “just attended the first Latin American screening of Emilia Perez and ooh, they ate that” or “She ate that. Best Saturday Night Live host ever.” That’s Ariana Grande. “Y’all ate that” or “She knows she ate that”, which means did a great job.
BEN: Do good.
HEDVIG: This is also…
BEN: Done good.
HEDVIG: This is fun that it’s in, but it is from Afro American English and/or drag shows.
BEN: I was just about to say, if it doesn’t come from the Black community, it comes from the gay community and I think that’s where this comes from.
HEDVIG: And I don’t know which one of those it is or if it’s somehow both. People can be Black drag queens. Woo!
BEN: My suspicion is that it’s LBGTQIA. I think it’s old school… Like you said, drag, queer, that sort of space. And one thing I’m not seeing mentioned here, which is a bit surprising, I would have thought that one of us would have acknowledged that this is sort of paired with SERVE.
HEDVIG: Yes, SERVE…
DANIEL: Interesting.
HEDVIG: …SLAY and a little bit like READ. Right?
BEN: Yeah. SERVE in particular, I think, because you serve things that then get eaten. I feel like there’s a genuine conceptual semantic link there. She’s serving. I won’t say the word here because it’s too much even for podcasts, but she’s serving…
DANIEL: We did that one without you: SERVING CUNT.
BEN: Oh, okay.
DANIEL: Yeah, yeah. We totally did that.
BEN: There we go.
DANIEL: So SANEWASHING, W and L, possibly AURA, BLUE ZONE and someone ATE THAT: our Words of the Week. Let’s get to some comments from Scott via email, hello@becauselanguage.com. Scott says, “I just finished listening to your dog whistles episode. I wanted to add what I think is a dog whistle that I’ve recently noticed. Seems to me when politicians on the right talk about LAW AND ORDER, they are using a dog whistle.”
HEDVIG: Yeah.
BEN: Oh, yeah. Old school. LAW AND ORDER has been a dog whistle for like 60 years.
DANIEL: Yeah. Where did you first notice it?
BEN: I grew up with LAW AND ORDER, like we all did.
DANIEL: Yeah. Like, Reagan especially.
HEDVIG: But what is it a dog whistle for? Do you want to explain?
BEN: Oh, policing poor people that we don’t like. In America, predominantly Black people. And I think that’s probably where it got its start. But America is far from the only place that is using the political concept of being tough on crime/having a law-and-order approach as like, “I’m going to make sure these poor people stay in their goddamn place.”
DANIEL: Yep.
HEDVIG: And to be clear, what it could mean… if you’re listening to this and you’re thinking like, “Obviously, that’s what law and order means,” it could mean things like policing financial crime. It could mean things like putting domestic abusers in jail. It could mean a lot of other things that…
DANIEL: Could mean the US Department of Justice being tough on a certain orange felon.
HEDVIG: …that the law applies to.
BEN: It has never been. [LAUGHS]
HEDVIG: Yeah, yeah, it could mean lot of things. And never does.
DANIEL: But it doesn’t. It never seems to mean that. Because if you’re in the in group, everything you do is great. And if you’re in the out group, everything you do is shit, so.
BEN: Just on this, I’ve been reading a fascinating book on prison abolition called “We Do This ‘Til We Free Us”. And in that book, they talk about the idea of, we could build a society in which we don’t require prisons. Everyone, when we advocate for the abolition of prisons and police and jails and all that sort of thing, say, “Well, what are we going to do with all the bad people?” And our answer to that is, how do we change society so that we stop creating…
HEDVIG: Bad people.
BEN: …this huge body of people who are harming society? And law and order could also mean that. Order as a word could mean: what programs could we institute to make sure that teenagers have places that they can express themselves safely and be part of community and all that sort of stuff? But it always means: how do we get police arresting more people on the street?
DANIEL: It always means: crime is the fault of people of colour. And that term goes back to Richard Nixon back in 1968, who gave a speech on the use of television. He took on the term LAW AND ORDER. Got picked up by Reagan, got picked up by lots of presidents since. A dog whistle I’ve noticed, by the way — just me — IMMIGRANT lately, which means person responsible for crime or somebody who US Republicans think should be deported. Whether they’re legal or illegal immigrants, they don’t seem to care. And I, as an Australian, do call myself an immigrant. I avoid the term EXPAT because I don’t want to say, because I’m one of the good ones.
BEN: Because weirdly, only white people get to be expats. [LAUGHS]
DANIEL: Yep. Thanks, Scott, for that observation. Big thanks to Dr Julie Sedivy. Thanks to everybody who gave suggestions for this episode, especially Gordon, MVP for this episode. Thanks to SpeechDocs for transcribing all the words. And thanks to patrons who keep the show going. And Ben and Hedvig, it’s great to see you again. Thank you for all you bring to the show.
HEDVIG: Thank you.
DANIEL: If you like the show, there’s some stuff you can do to help us. You can follow us, we’re @becauselangpod on just about every conceivable social platform. Lots of people have sent us ideas via our Discord or email, hello@becauselanguage.com or via SpeakPipe. Did I say SpeakPipe twice? I don’t know.
HEDVIG: [LAUGHS]
DANIEL: But you can put… [CROSSTALK]
BEN: No, maybe.
HEDVIG: No, there’s no way. There’s no way.
DANIEL: I guess not.
BEN: Yeah.
DANIEL: Okay.
BEN: It’s out there. It’s in the stone tablets.
DANIEL: Can’t go back. Can’t go back. You can also tell a friend about us or write us a review and we have a new review from Nati. You might remember Nati from our Episode 105 or 500, who joined up. He gave us some great stuff. This is the review from Nati. “Best Sci Coms on the Internet, five stars. Now that I am out of college, but pre-grad school, I found it hard to keep up with the state of the field. I felt as if y’all were heaven sent when I stumbled upon your podcast. Y’all are amazing science communicators, give real times update on the latest science and are overall hilarious.” Obviously, that’s not me. “Also, y’all have awesome politics.” That is me. “It’s like we can be friends?? Anyway, I feel like I just learned so much from y’all as a budding linguist, educator and science communicator myself. As I like to tell my students: are you Mario when it comes to turts? Cuz you’re crushing it.”
[LAUGHTER]
BEN: I was not knowing where that was going. That one tiny sentence was one wild ride. [LAUGHS]
DANIEL: BA-DOOP! “Absolute 10/10 cannot recommend enough.” Thanks, Nati. That was great. Everyone else, was that so hard?
BEN: Say that. Say more stuff like that.
HEDVIG: That’s very nice, Nati. As with a recent episode, I do struggle with too positive judgments. Can we get like five stars but people write shorter and less embarrassing texts?
DANIEL: Less laudatory?
BEN: What I want to do is get some of those five-star reviews and then the text is just hating on me specifically.
DANIEL: We’ve had those.
BEN: That’s what I want to hear. Just me.
DANIEL: We’ve had those. Chris did one.
BEN: I want to hear Hedvig compliment. I want to hear Daniel compliment. And then, I want someone just to lay into me. It’s going to be heaps of fun.
HEDVIG: Yeah. Exactly.
DANIEL: [LAUGHS] Sorry, you guys. But when you do awesome work, you’re going to get noticed. That’s just the way it’s going to be. Go, Ben.
BEN: Hey, do you know, Because Language is actually, if we’re being really honest at this point, mostly made up of our patrons. And what I mean by that is, patrons, obviously Daniel does a lot of good work, and Hedvig and I are just like along for the ride, riding them sweet, sweet coattails all the way to the top.
DANIEL: Very kind.
BEN: The reality is a lot of even what Daniel does is stuff that is sent to us by the patrons or articles that get provided to us by the patrons or things that the patrons come up with and they tell us about. We have this wonderful, wonderful, wonderful community of patrons. And if you would like to be a patron too, you can get on board. It not only helps us make the show, but it also helps do things like transcribe the show so that people who can’t hear can access the stuff that we talk about. And it helps us offer little humble stipends to the people who come on the show because even though they’re talking about things that they love, labour should not be free in this world and we think that it is important to offer them money. And a lot of them just donate that money to charity or get us to donate the money. But that’s good too, and your money could help with that.
Now, we like to read out our patrons and we like to change the order so it’s not hella boring. This time we’re ordering our supporters…
HEDVIG: Daniel likes to change the order.
DANIEL: [LAUGHS] I do.
BEN: I’m going to be honest. I kind of like it too. [DANIEL LAUGHS]
HEDVIG: Aright.
BEN: I’ve come around.
DANIEL: Okay.
BEN: This time, Daniel/Ben, through no labour or effort on his own part…
DANIEL: As usual.
BEN: Has ordered the supporters by the score their name would get in Words with Friends, divided by the score their name would get in Scrabble, absent any bonus squares.
DANIEL: They are different, you know.
BEN: So basically, how juicy is your name if you were to play the word games with it. So: favoured by Scrabble — so succeeds in Scrabble, but not so great with Words with Friends — the dearest of my heart Ayesha, Kathy, Keith, Whitney and PharaohKatt. And the same across games, so does well in both: Amy. Really? Amy? That short?
DANIEL: Yeah. The Y doesn’t seem to have much of a difference. Well, A, M, and Y are the same score in Scrabble and Words with Friends. Nothing different.
BEN: Rach, Tadhg, Kate, Rhian, Meredith, Andy, Tony, Alyssa, Chris W, Kristofer, Larry, Lyssa, Termy. How could Termy be at the end if Amy…? Anyway, anyway,
DANIEL: They’re all… yeah. I don’t know.
BEN: And favoured by Words with Friends: J0HNTR0Y, Cheyenne, Felicity, Margareth, Hedvig is here. What?
HEDVIG: That is very strange.
DANIEL: Hedvig, this is where you came in the score.
HEDVIG: Oh.
BEN: Okay. Hedvig sits at this point. Chris L, Canny Archer, Helen, Rodger, gramaryen, WolfDog, Ariaflame, Diego, Molly Dee, Andy from Logophilius, LordMortis, Amir, Steele, Kevin, Matt, O Tim, Jack, new this time: Aldo
DANIEL: Hi, Aldo.
BEN: Raina, Elías, Sæ̃m?
DANIEL: Mm-hmm. Sæ̃m.
BEN: Sonic Snejhoj. [BLOWS A RASPBERRY]
HEDVIG: Why didn’t I get higher?
BEN: Sonic Snejhog, Nikoli, James, aengryballs, Rene, Stan, Daniel is here.
DANIEL: Pretty low down.
BEN: Ignacio, Joanna, Nasrin. Ben is here.
DANIEL: This is for reference.
BEN: Oh god. Colleen, Nigel, Manú and Luis. And our newest patrons: At the Listener level: SCW and Eve. Eve started at the Friend level, but quickly upgraded to Listener. And our latest free patrons, because you can do that too, if you just want to be like, “Hey, I like you guys, but I don’t have any cash.” Tamara, Gregory, Jacqui, Jo, NMC, and Aayush. Thank you to all of our patrons. I hoped you liked that particularly novel ranking.
HEDVIG: Our theme music was written and performed by Drew Krapljanov, who also performs with Ryan Beno and Didion’s Bible. Thanks for listening, and we’ll catch you next time. Because Language.
IN UNISON: Pew, pew, pew.
[BOOP]
DANIEL: Doug and Trudy did stickers for us.
BEN: You showed me, it’s great.
DANIEL: Linguistic chaos goblins. Yep.
BEN: Yeah, yeah, yeah.
HEDVIG: Very cute.
DANIEL: I like them a lot. They’re from Oglaf. So, that’s for our patrons.
HEDVIG: Daniel, did you read the webcomic, Oglaf?
DANIEL: I am a huge fan of Oglaf.
HEDVIG: Okay. All right. Okay.
DANIEL: I’ve been their patron for a long time.
HEDVIG: Oh, okay. It just seemed…
DANIEL: What?
HEDVIG: I don’t know, more D&D and saucy than I thought you were into.
BEN: This is one of the strange things about Daniel, right? He’s a man of contradictions. And as nearly all things…
DANIEL: What’s the contradiction?
BEN: …we can relate this back to Mormonism. Here’s Ben’s grand unifying Daniel impression of Mormonism. Okay, here’s how this goes. This is my belief. It could be completely wrong. Daniel spent a lot of his life being told that the saucier, funner, spicier part of life is fucking gross and evil and weird and you should only ever possibly talk about it with very close people and all that kind of stuff.
DANIEL: You’re doing well so far.
BEN: And then, as a thinking person, he went out into the world and he basically went, “Okay, so pretty much everything that they said was bad is suspect. So, now I’m going to take myself through a bunch of those things. Right? And basically, I get to choose.” And some of the stuff, like alcohol…
DANIEL: I get to choose.
BEN: …he was like, “Meh,” take it or leave it. [DANIEL LAUGHS] But understandably…
DANIEL: But sex!
BEN: He got to sex, and he was like…
DANIEL: Oh, my god.
BEN: “Oh, guys, you really buried the lede on this one.” [LAUGHS]
DANIEL: This in my zone! [LAUGHS] Oh, I was kind of into sex already; I just felt really bad about it.
BEN: Yeah, yeah. Whereas Oglaf is really celebratory and really sex positive. And it’s a very different approach to sexuality than I think Daniel probably grew up around. [LAUGHS]
[BOOP]
BEN: Look, it’s Sandy.
DANIEL: It’s Sandy.
HEDVIG: It’s Sandy.
BEN: He’s a pretty boy.
HEDVIG: He’s a pretty boy and he wants a lot of attention.
BEN: He’s getting fluffier. Is that because it’s getting colder or is he just getting fatter?
HEDVIG: I think also he’s gaining a bit of weight.
[LAUGHTER]
DANIEL: Very good.
HEDVIG: But they are extremely cuddly as it gets colder. They like to sleep on top of me all night.
BEN: Yeah, yeah, I’m the same way. But I get really torn because we’re warming up now and neither of the cats are sleeping on our bed anymore. And I’m like, “Come back.” As much as… But then when they’re sleeping with me and I’ve got two cats in the V of my leg preventing me from rolling over, I’m like, [ANGRILY] “Get off.”
HEDVIG: Yes. Yes.
DANIEL: Stupid felines.
HEDVIG: I have just decided that my ability to move is valuable and sometimes they have to scooch.
[LAUGHTER]
DANIEL: That’s not very Buddhist.
[BOOP]
BEN: It was an amazing trip, and I had an amazing time. And I strenuously recommend to anyone out there who is considering doing a long hike or a long bike trip or something like that, and they’re like, “Oh, I’d love to, but I don’t know. Blah, blah, blah.” I’m here to tell you, just do it. And not like in the weird Nike, like, be the Adonis sense. But I did no training. All you’ve got to do is just keep going. And so much of these really long format expeditions are not about being super fit or whatever. It’s just about being like, “Tra, la, la, la, la, la. Oh, look, a nice lake. I’ll stop here.” [LAUGHS] That was me, times like 49 days.
HEDVIG: Nice.
DANIEL: That’s a brilliant lesson.
BEN: So, yeah, the closest thing to spirituality I found on my trip was that significant a departure from the rigmarole of a capitalistic, sort of grindy life really does allow you to access a different sort of state of just being a human. And that’s really, really lovely. It’s so lovely to just spend a couple of weeks. Because for me, it took a couple of weeks to get out of the mode of like, “Am I riding far enough? Am I being productive enough?” Like, blah, blah, blah.” And then, just being able to be without goals and KPIs and all that sort of stuff, that’s really rare in our world and really refreshing. So, that’s my sales pitch.
HEDVIG: That’s really nice.
BEN: So, I should acknowledge the privilege of having a healthy and capable body that was able to keep going every day and all that sort of thing.
DANIEL: Well, it is nice to have you back. I don’t know if you know this, but I took a trip to the States that nested inside your trip to the States.
BEN: What, and you didn’t come and see me?
DANIEL: I thought that would be the funniest thing ever to find out where you were going to go past, and just stand there as you’re whizzing by on your bicycle, and be like, “Hi, Ben!”
BEN: I swear to god, if you went to Spokane… Did you go to Spokane?
DANIEL: Yeah, I totally went to Spokane.
BEN: You motherfucker. I was so close to there.
DANIEL: [LAUGHS] Don’t worry. I knew I’d see you when I got back.
BEN: I was in Western Montana, you prick.
DANIEL: Well, I knew that I was far away on a bicycle. I wasn’t going to go… Well, I could have taken the car.
BEN: Yeah, you really could have! [LAUGHTER] It’s just around the corner. Anyway, anyway.
DANIEL: I would have had to track you down.
BEN: I still love you.
[BOOP]
DANIEL: My two young ones have started me on the Spotto game, where you have to spot a yellow car and say Spotto. And it’s been really interesting as a colour thing because I’ll see one that’s kind of yellowy green and I’ll be like, “Spotto?” And they’ll be like, “No,” and there’ll be a burnt orange, and I’ll be like, “Spotto?” And they’ll be like, “Yes.” And I’ll be like, “That’s not a Spotto. What are you talking about?”
BEN: [LAUGHS] There needs to be consistency!
HEDVIG: In Sweden, there’s a popular game that if you see a yellow car, you get to punch someone else.
DANIEL: Yes.
BEN: This is Spotto. We are talking about the same game…
HEDVIG: Oh, okay.
DANIEL: Used to be Volkswagen Beetles.
BEN: Except Daniel presumably is playing it nonviolently because he is, like, a nice human father.
HEDVIG: So, in many families that I’ve been to, there’s a rule that if the car is like a postal car and therefore is yellow, that doesn’t count.
DANIEL: Yes, that’s what I say. And they don’t believe me. I say if it’s obligatorily yellow, then it doesn’t count. Thank you, Hedvig. In your face, kids.
[Transcript provided by SpeechDocs Podcast Transcription]