What one thing explains the most about language?
Here to answer that question is linguist and friend of the pod Dr Mark Ellison — and many of our listeners and friends. It’s one of our Deck Chats! Grab a coffee and follow along.
Timestamps
Cold open: 0:00
Intros: 0:31
Mark’s work: 1:51
Listener responses: 9:07
Daniel’s answer: 43:34
Mark’s answer: 1:03:45
Wrap up: 1:17:17
Video and audio
Watch this episode
Listen to this episode
Bonus video
More on topological structure
Patreon supporters
Thanks to all our patrons! Here are our patrons at the Supporter level.
- Whitney
- Matt
- Chris L
- Termy
- Helen
- Jack
- PharaohKatt
- LordMortis
- Lyssa
- Elías
- gramaryen
- Larry
- Rene
- Kristofer
- Andy
- James
- Nigel
- Meredith
- Kate
- Nasrin
- Joanna
- Nikoli
- Keith
- Ayesha
- Steele
- Margareth
- Manú
- Diego
- Ariaflame
- Rodger
- Rhian
- Colleen
- Ignacio
- Sonic Snejhog
- Kevin
- Andy from Logophilius
- Stan
- Kathy
- Rach
- Felicity
- Amir
- Canny Archer
- O Tim
- Alyssa
- Chris W
- aengryballs
- Tadhg
- Luis
- Tony
- Wolfdog
- Molly Dee
- J0HNTR0Y
- sæ̃m
- Linguistic C̷̛̤̰̳͉̺͕̋̚̚͠h̸͈̪̤͇̥͛͂a̶̡̢̛͕̰͈͗͋̐̚o̷̟̹͈̞̔̊͆͑͒̃s̵̍̒̊̈́̚̚ͅ
- Amy
- and Aldo
Become a Patreon supporter yourself and get access to bonus episodes and more!
Become a Patron!Show notes
Emphatic Pronouns | Scottish Gaelic Grammar Wiki
https://gaelicgrammar.org/~gaelic/mediawiki/index.php/Emphatic_Pronouns
Don’t Count on It | Scientific American
https://www.scientificamerican.com/article/dont-count-on-it/
The Interpreter | The New Yorker
https://www.newyorker.com/magazine/2007/04/16/the-interpreter-2
Brainfuck | Wikipedia
https://en.wikipedia.org/wiki/Brainfuck
How Deaf Children in Nicaragua Created a New Language
https://www.atlasobscura.com/articles/what-is-nicaraguan-sign-language
T. Mark Ellison & Uta Reinöhl: Compositionality, Metaphor, and the Evolution of Language
https://link.springer.com/article/10.1007/s10764-022-00315-w
POWERHOUSE COLLECTION
https://collection.powerhouse.com.au/object/306154
Transcript
[Transcript provided by SpeechDocs Podcast Transcription]
DANIEL: This is going to be huge. I better take a sip of water.
MARK: [LAUGHS] I need to fill myself up with some coffee for the grandness of it.
DANIEL: Listeners, please do the same if you need to. It’s going to be big.
MARK: Pull your socks up tight.
[BECAUSE LANGUAGE THEME]
DANIEL: I’m here with Dr Mark Ellison for one of our Deck Chats. Hello, Mark.
MARK: Hi. So good to be back here for another Deck Chat.
DANIEL: It’s a very special episode of Because Language because not only is it one of our deck chats, it’s also a further episode with you at my place, which is always fun.
MARK: It is always fun.
DANIEL: Last time, the weather was different for our Lazy in a Good Way episode.
MARK: It was late November, which was possibly part of the reason why the weather was different. Not the middle of winter.
DANIEL: Yes, we’re in the middle of winter right now.
MARK: Yeah.
DANIEL: And it’s a cold, sloppy day.
MARK: It’s a very damp day.
DANIEL: It’s damp.
MARK: Been saying “damp” a lot today.
DANIEL: Just damp. Why do we do this to ourselves?
MARK: I don’t know. I was just actually thinking, because we like words. I would mention a Scottish word for this sort of weather.
DANIEL: Oh, very good.
MARK: Dreich.
DANIEL: Sorry. Breich?
MARK: Dreich.
DANIEL: D-R
MARK: E-I-C-H.
MARK and DANIEL: Dreich.
MARK: That’s how I spell it anyway.
DANIEL: And what does… That’s your headcanon. What does Dreich mean?
MARK: Basically, it means damp, grey. It’s like little is going for it for this day.
DANIEL: Okay.
MARK: It’s like you probably don’t want to be outside.
DANIEL: But we are outside because…
MARK: We are outside.
DANIEL: …this is a Deck Chat. And we’ve got coffee.
MARK: And we’ve got coffee.
DANIEL: I remember we talked about prominence. That is to say, making something more oomphy. That’s the word I’m using. Giving it more oomph.
MARK: Yep.
DANIEL: To attract the attention of a listener that may be wandering off or to just make something more salient, making it more big so that you can correct mistaken impressions or so that you can… It actually helps memory…
MARK: Yes.
DANIEL: …as with strings of numbers.
MARK: Yep.
DANIEL: So that’s all…
MARK: That’s all true, but that’s not all of prominence.
DANIEL: Okay.
MARK: So, that is what we call code prominence, or possibly signal prominence.
DANIEL: Okay.
MARK: Signal prominence is prominence that happens just automatically because of the nature of the way something’s expressed. Like, if somebody says something very loudly to you, shouts it, that would be more prominent just because of natural properties of the sound.
DANIEL: And then I would say, “Oh, that seems relevant. They must have done that for some kind of reason.”
MARK: Exactly. So, what you’re doing then is you’re taking that as a cue to what we call a predictive prominence or discourse prominence, which is your expectations about what’s going to happen.
DANIEL: Mm.
MARK: And we can think of that as kind of an internal kind of prominence, where as opposed to the external kind of prominence. And the internal prominence is how likely you think something is to crop up again. It could be how likely a particular word is to crop up again, particular concept or a particular entity in the discussion that’s happening. And so, we have this… or even phoneme. I mean, it’s all these.
DANIEL: Wow.
MARK: We make predictions about everything all the time.
DANIEL: Wow. Okay.
MARK: And the more prominent something is in terms of our internal state, the higher the expectation we’re giving it.
DANIEL: Okay.
MARK: But what I learned that’s new…
DANIEL: Yes. Okay. I wanted to go off on that.
MARK: Oh, yes, let’s off on that.
DANIEL: I wanted to go off on that a little bit because it reminded me of something. We’ve been talking about Stephen Levinson’s work on the show a lot. Hedvig and I are both reading a lot of that. And the way Hedvig expressed it was, “If you say something complicated, you mean something complicated. And if you say something normal, you mean something normal.” So, if I say, “That’s not my ferret,” then I just mean that’s not my ferret. But if I say, “That’s not my ferret,” now I said that in a different way, and now I mean something different.
MARK: I think a lot of the things that I will be talking about from my side today are very connected with this.
DANIEL: Okay. Good. I’m glad [CROSSTALK]
MARK: I think we’re on the same page. But I would say that in this particular case, often what we’re doing, like where you’ve used this kind of prominence, put extra prominence on the part of that expression, what you’re doing is you’re saying something like, “You might believe that’s my ferret, but it’s not my ferret.”
DANIEL: Yeah.
MARK: So, you’re saying perhaps you have a false belief, or for some reason I need to put extra effort in to push the attention to what I’m saying up higher because you have some false belief and your belief that it’s not my ferret, is how you have that is very low probability because you think it is my ferret, and I have to push that up. Do lots of work to push that up high.
DANIEL: So, that then you predict a thing, a scenario in which it’s really, really not my ferret.
MARK: Exactly. So, you’re basically shifting a belief from one direction to the other direction.
DANIEL: Did you say 4-5-6? No, I said 4-NINE-6.
MARK: Exactly. So, somebody’s got a mistaken idea and you’re correcting it. That involves more work than giving them information from they have no idea.
DANIEL: Mm-hmm. Okay, good so far.
MARK: Yep. So, one of the things that now we’ve got this kind of internal prominence, the predictive prominence versus the expressive prominence, where we have that code prominence as well, which is kind of conventional expressions of prominence. So, for example, in Scottish Gaelic, we can make pronouns prominent. The way you would do that in English is by stressing them. So, make them use a bit of signal prominence, make them a bit louder and pitch a bit higher and things. But in Scottish Gaelic, you put an ending on them, and that’s conventional. And it’s like you… one of the things it’s hard to learn when you’re learning that is not to stress in the English way, these pronouns, because that’s not what you do in Gaelic.
DANIEL: So, you would stress it but you wouldn’t stress it by making it louder, you would just say it at the same volume, same pitch, and add an ending, and that does the job.
MARK: It does the job. Maybe you do change the way you say it, but the main thing is putting the ending on, and it’s perfectly fine if you don’t change how you’re saying it.
DANIEL: Okay.
MARK: Okay. So, that’s a different kind of… We call that code prominence because it’s part of a conventionalised code.
DANIEL: Instead of signal prominence.
MARK: Which is just like a natural body reaction to something…
DANIEL: Okay.
MARK: …like a rising pitch.
DANIEL: Yep.
MARK: Sounds like something that’s fast approaching us so it will attract our attention.
DANIEL: Got it.
MARK: Yeah. So, the thing that I discovered from the discussions that we had in our group was a puzzle I’d had, which was what one of our groups was doing was looking at things like the prominence associated with when we’re referring to people rather than referring to objects or where we’re referring to agents, like active participants in a particular action versus the patient, the thing that has the action done to it. I didn’t know where to fit this into the puzzle.
And during these discussions, I kind of realised, “Ah, it’s about, if I don’t know anything about the context, just get this sentence or this concept presented to me, how much prominence would I assign to it in terms of predictive prominence? So, how likely is this thing to be talked about again, just because of what it is?” And one of the things is: we like talking about people more than we like talking about things. So, if I’m referring to a person, it’s more likely to come up again than if I mentioned, “I saw a stick in my backyard.”
DANIEL: Yeah. “I had some cheesecake.”
MARK: Yeah. [LAUGHS]
DANIEL: The cheesecake is gone, and the people remain.
MARK: [LAUGHS] Exactly. It’s true.
DANIEL: Okay.
MARK: Yeah. So, that was like a light bulb moment for me that came out of this process of which seemed like unrewarding in some ways. Obviously, apart from four years, nothing, but I realised, “Oh, actually, this is really useful. We talk to each other in different ways as we’re doing this process,” and I think it’s making me feel a little bit more positive about this grant application stage of academia that most academics feel is like a necessary evil rather than something that can be rewarding.
DANIEL: Yeah. Wow. And yeah, you managed to make it work for you, and you’re working together to build knowledge.
MARK: That’s it. That’s what the whole game is about.
DANIEL: Cool. Thank you. Well, for today’s discussion, we wanted to have a chat out on the deck, but we didn’t know what we were going to talk about.
MARK: Mm-hmm.
DANIEL: We figured we’d come up with something.
MARK: Yeah.
DANIEL: And then while were kicking it around, I had this idea, and it’s one simple question, and we challenged each other to answer it.
MARK: Yeah.
DANIEL: And the question is: what one thing explains the most about language?
MARK: Yes. That was the challenge.
DANIEL: What one thing explains the most about language? And I put the question to our Discordians and to everybody on our social networks, and they came back with a lot of answers.
MARK: I’m so excited to hear about this.
DANIEL: Okay, here we go. This one… First one’s from Diego on Discord.
MARK: Yeah.
DANIEL: His one thing is: “Any language can express any idea or concept – all languages are equal and valid!” That’s super good. That’s a super important principle. Let’s just talk about that for a second, even if you don’t feel like it’s the one big thing.
MARK: No, I think that’s a great thing.
DANIEL: It’s a great thing. It’s kind of weird that there are no languages where a concept is just unavailable. You might have to circumlocute. You might have to try and struggle a little bit, but there are no languages where it’s like, “Oh, no, you can’t get there from here.” Like the New England gentleman. Every language can do all the things.
MARK: That is a great question. And I wonder: do we know that for sure?
DANIEL: Okay, well, I guess I would challenge somebody to give me a counterexample. What’s a concept that simply can’t be translated?
MARK: Okay, so I’m thinking about — and you’ll be able to correct me because I think you’ve read this material more than I have — about the Pirahã from Daniel Everett and numeracy.
DANIEL: Yes.
MARK: It’s like if somebody’s not doing… doesn’t have the concept of addition, how do you get from a small set of numbers, which they might have, to 134?
DANIEL: Well, it’s a funny thing about Pirahã. They have numbers now. They just went ahead and borrowed them from Portuguese.
MARK: Yeah, this is fine. This is great. But I would say that what they’ve done is they’ve changed their language. So, I think that they’ve realised they had a gap in their language. Their language wasn’t good for everything.
DANIEL: Yep.
MARK: But I think there’s an easy fix, and I think if as long as you accept that there’s an easy fix for a language by borrowing stuff from your neighbours…
DANIEL: You might have to engage in some pretty wild shenanigans like circumlocution, like I said, or just borrowing entire concepts.
MARK: Yeah.
DANIEL: But you could do it, I wonder, in pre-Portuguese numbers, could you do that in Pirahã? You’d have to sit down with someone. And in fact, I do remember Dan Everett talking about how they said, “Oh, no, we don’t want to do that.” So, there was a cultural barrier to them counting.
MARK: Exactly. So, I think that there are… And I think… and this still comes back to this point, which I’ll probably take up in a later discussion as well. So, I think in general, this point is really true and really interesting. But what’s cool is I think some languages make some things harder. Other languages make other things harder.
DANIEL: We’ve talked about this.
MARK: Yeah, exactly. Because when we’ve talked about laziness, it’s like some kinds of laziness is telling us about what a language cares about.
DANIEL: Yep. So, for example, on computers, if I were doing lots of work with text and regular expressions, I would use Perl. I wouldn’t use Java because it’s not easy in Java.
MARK: Exactly. The thought puts my teeth on edge. [LAUGHS]
DANIEL: Perl is just better for that. Or Python or something.
MARK: So much better. So much better.
DANIEL: So, you think that’s true for natural languages?
MARK: I think that’s true for natural languages as well. And I think, for example… It’s hard to think of because we have so much of a global culture nowadays that lots of things come together. But here’s a nice example. When I was learning Gamilaraay in ANU, it was the only aboriginal language which was taught at university, as a university course in Australia. I did that just because it was really fun and available. But as I studied more of this, I found they had this cute thing where you can inflect verbs for time of day.
DANIEL: Inflecting verbs for time of day.
MARK: So, it means you can put a different ending on the verb that just marks, did this thing happen in the morning or in the afternoon?
DANIEL: So, in English, that would be like, you just say, “I ate,” and we can modify that with I ate lunch. But if there were little, tiny particles for, like, “eb” and “el” and “ed” for breakfast, lunch, and dinner, I could say, “I ate-eb.”
MARK: Yeah, exactly.
DANIEL: That’s not time of day. That’s meal, but okay.
MARK: Yeah, but if it was time of day, like if it was morning, then you think, oh, it’s probably breakfast. It was afternoon, maybe lunch. If it was night, dinner.
DANIEL: Yeah.
MARK: So, what’s really cool is that… And these then are separate from the tense markers. So, if I use it with the morning one with the future marker, it’s tomorrow morning.
DANIEL: Mm-hmm.
MARK: So, I can easily say, just without extra words, just with the inflection on the verb, I can say, “I will do something. I will talk to you tomorrow morning.” The tomorrow morning part just is part of the verb.
DANIEL: That’s a language that’s set up for that thing.
MARK: It’s set up for that thing, for making this distinction and making it often and easily and clearly… And you just think, “That’s really nice, because I wouldn’t think about that in English.” We always think, “Oh, it’s like an add-on I might have, but I might usually would leave out because it’s too much effort for the purpose.”
DANIEL: Yep. Okay. So, there are actually two things in Diego’s answer. We will get past these. We will get to all these.
MARK: Sure.
DANIEL: Any language can express any idea or concept. So, that’s one thing. But then, it has an interesting philosophical outgrowth, and that is, all languages are equal and valid, which is nice from the perspective of decolonisation or not being such a terrible prescriptivist in your personal life or whatever. It’s got implications.
MARK: I don’t know that that follows, basically.
DANIEL: Oh, okay.
MARK: Let me give you an example, as we’ve talked about before with computer languages.
DANIEL: Mm-hmm.
MARK: Okay. I don’t know if I can say the name of this for our wider audience, but there is a computer language that’s called Brainfuck.
DANIEL: Okay.
MARK: And basically, it’s designed to be as horrible as possible, and it really is. It succeeded. It is really, really, really horrible.
DANIEL: Wow, okay.
MARK: Having as many of the most… If I remember rightly, most characters in this language are irrelevant, but the ones that are significant are very similar to each other. It’s like an…
DANIEL: [LAUGHS] Oh, no. Okay.
MARK: O and 1. Capital O and 0 and things like that to try and mess with your head as much as possible. And you really don’t want to spend much time with this language. But it’s Turing complete. It can express everything every other programming language can express.
DANIEL: And it’s valid for certain purposes if your purpose is to have a terrible time and have fun trying to code.
MARK: Yeah. To write code nobody could possibly read or understand, which was the intent when it was invented.
DANIEL: Okay.
MARK: But the point is, theoretically, it’s possible to have a language which is very unpleasant to use.
DANIEL: Yep. Okay.
MARK: That is just as expressive as a happy, lovely language like Python or Haskell.
DANIEL: But in what ways are these not equal? They are…
MARK: They’re equal in their… And I think this is going to be important for our wider understanding of language that we’ll talk about today, and that is that we have expressiveness in the sense of, “I can express this by putting in up to infinite amounts of effort. With enough effort, I can express this thing,” which we talked about in terms of, like, even if my effort involves borrowing from another language, I can do it in this language, kind of. Yeah. And that’s different from, “It’s easy to do in this language.”
And I think this is a really important distinction that we should have, is this kind of not thinking about these things in black and white makes them more useful. I agree, there’s this sense of we’ve decided culturally and socially for good reasons, because of terrible things in history, that it’s good to say all people are equal in terms of their rights. And for example, right to life, we would very much like to see a world in which everybody is treated equally, but that doesn’t mean that everybody is equally capable of doing particular things.
And it’s the same with languages. For some, just as for me, writing an essay in Navajo would be practically impossible. Although theoretically, if my life had been different, it could have been entirely easy. The same is true for languages, if their histories have been different, they could have grown up in a different way and had different facilities. But that doesn’t mean that they have… Everything is easy for them equally.
DANIEL: There’s no natural language like Brainfuck.
MARK: No, because those get pruned very quickly in use. I think some languages… Let me give a quick example.
DANIEL: And then, we will move on.
MARK: We will move on. The Nicaraguan Sign Language.
DANIEL: Yes.
MARK: Starts out a little bit more like Brainfuck, because people are like…
DANIEL: Now, Nicaraguan Sign Language was the one. We talked about this with Adam Schembri right here on the deck. It’s the one where the parents sent deaf kids to a school, but the school was using Oracy. So, they were trying to teach kids to lip read basically, no sign language. The kids, on their own generated a sign language, and that’s now Nicaraguan Sign. And it went through iterations like there was a first generation and then there was a second generation and more.
MARK: This is exactly the point I wanted to get to, is that first generation, I’m not saying it was like Brainfuck. It wasn’t a smooth…
DANIEL: A bit janky.
MARK: It was a bit janky. It was a bit like…
DANIEL: Hadn’t had the rough edges knocked off.
MARK: Exactly. It was like driving a 1980s Volkswagen sedan, not Beetle. It was just like… I had one of those, and I wished it would die so I could buy a car that ran smoothly. And I think what we see within the Nicaraguan Sign Language is that the children who grew up learning this from a very young age because their peers had developed this sign language, but then the younger ones grew up becoming native speakers of this thing, they do it smoothly and quickly, and it’s really fast, and it’s very expressive in a way…
DANIEL: Efficiently.
MARK: Exactly.
DANIEL: They made it efficient.
MARK: In human terms of efficiency, which is really a beautiful smoothness. It’s like if you watch an amazing dance, it’s got that human efficiency in motion. And that’s what we have in natural languages. But it doesn’t take long for us to get there from wherever we start, we can go in one generation to something that’s awkward and negotiated, because these individual children who created this language, they didn’t have something to build on. They were creating this from scratch. And so, their first go was not perfectly smooth and unified, but it just took less than a generation to become a smooth, fully functioning human language. We’re really cool at that.
DANIEL: Well, this takes us to the next one, the second one from The Fishmonger on Patreon. Thanks, Fishmonger. Their one fact is that isolated deaf communities with minimal shared signs spontaneously create grammar.
MARK: Yeah.
DANIEL: It’s just one of these things about the human language impulse. Like, with Nicaraguan Sign, we just do it.
MARK: Yep.
DANIEL: And I feel like we don’t need to discuss much more of that.
MARK: I would add one little thing, if I can.
DANIEL: Okay.
MARK: Which is, what’s the alternative to having grammar?
DANIEL: The alternative to having grammar would be putting words in random order, and that would impose far too much work on the processing side.
MARK: Yes.
DANIEL: It would be very nice for the signer. You could just do whatever. And in fact, in small communities where everyone can sort of know everyone and memorise each other’s quirks, word order can be a bit freer.
MARK: But it’s not as free as you think. It’s still controlled, I would argue, by the salience we see in these free word order languages. It’s free grammatically, but it’s not free in terms of what we call the information structure, which is how predictable particular things are, so you tend to get the predictable information first and the less predictable information coming after.
DANIEL: Going from old to new.
MARK: Exactly. And this is part of, when I’m talking to you, you don’t know exactly what I’m going to say next, and so I start with something that’s going to be easy for you to grab, and then we can build on that in terms of expectations.
DANIEL: Well, that’s what we do with stories. We say, “Once upon a time,” that’s how you got a story. You’ve got the story built up, so you know that’s how the story.
MARK: My brain is thinking…
DANIEL: That’s right.
MARK: …princesses and dragons and castles and…
DANIEL: Okay, so now I’m going to add something new. Once upon a time, there was a castle.
MARK: Okay.
DANIEL: That was a new thing, but it’s old now.
MARK: It’s old now.
DANIEL: And this castle, it’s old. And this castle, that’s why I start with the old thing, right? This castle was on the banks of a beautiful river. It, the river…
MARK: Yeah.
DANIEL: …flowed to the ocean. And in that castle lived…
MARK: Exactly.
DANIEL: We’re always going from old to new.
MARK: Exactly. And a friend of mine in Cologne, as part of her PhD, spent quite a bit of time looking at topics and topic extensions. It’s a longstanding idea, but she has this nice framework for looking at these things where there’s an extension to an existing topic, and that’s treated as part of the known information because it’s kind of expected on the basis of what was explicitly mentioned beforehand.
DANIEL: Okay, let’s go to aengryballs on Discord, whose one big fact is: regardless of whether universal grammar is a thing, you have to learn language from other humans in order to have language. You aren’t born a speaker, and you can’t learn it independently. That’s interesting, isn’t it?
MARK: That is so cool. I like that. It’s like saying it’s not innate.
DANIEL: It’s not innate. I mean, we have bodies that are very well set up for language.
MARK: Sure.
DANIEL: We have the vocal tracts we do. But that’s just a long process, where conversation, communication sculpted us, and tool making sculpted us, and lots of things in our society and our brains and our bodies and our communication worked together to build who we are.
MARK: That’s one of the things that I think previous guests of yours, Nick Chater and Morten Christiansen, in their book, they…
DANIEL: The Language Game.
MARK: The Language Game. They made this great point that language seems much newer than our biology. So, probably language evolved to fit us rather than us evolving to fit language but…
DANIEL: I’m going to… That is getting into my big point.
MARK: Okay.
DANIEL: Which we haven’t even gotten to yet.
MARK: Okay, so let’s get back to this point. This point is about the fact that we have to learn language from other people. I think what we see in not just humans, but throughout the least vertebrates, and I think probably this is true of insects and other hard-shelled things, that our neural mechanisms are very plastic. And it’s so much easier, not just for us, but for evolution, to dump as much of the information that a creature needs in the world and use general purpose machinery to just learn the context that it lives in that’s, A, more robust in terms of evolution. Like, if the environment changes, well, it’s built in adaptation it can cope. And that’s very helpful.
And what we found, for example, is something that seemed we thought might be hardwired, was that in the back of your brain, this is a bit that processes visual information pretty much directly from the eyes the visual cortex. And what we find there in humans is there’s these stripes of information from the left eye, and then there’s a stripe of information that from a similar stripe in the right eye that would be looking most of the time roughly the same thing. And then, a stripe from the left eye is a little bit over, and then the stripe from the right eye that matches, most likely, they have the same information. You’d be looking at the same thing, given that our eyes can wiggle, so it’s not always exact. We can have our eyes focus further apart or closer together, depending on how near the object is we’re looking at. The point is we thought, and it was kind of natural to think that was a hardwired fact about binocular vision versus animals that have their eyes pointing in different directions.
DANIEL: Like deer.
MARK: Like deer or frogs that have their eyes growing out the sides.
DANIEL: Those weird lizards.
MARK: Those weird lizards, chameleon lizards, they’re so cool.
DANIEL: Yeah.
MARK: But people did experiments that I would be squeamish about doing. But people did these experiments where in a tadpole, they implant an extra eye bud, like the bit that goes into an eye next to an eye that was existing. So, this frog had one eye one side and two eyes on the other side. And they got exactly the same organisation in the brain from the eyes that had overlapping fields of vision while they had a separate… you don’t have these stripings with the eye that was looking in the other direction. And so, what this seems is it’s just a plastic response that animal that had for a long time has not had binocular vision if ever its ancestors had it…
DANIEL: It got it.
MARK: …you put the eyes close together, they get the processing they need.
DANIEL: They weren’t evolutionarily set up for binocular vision, but their bodies just used what it had.
MARK: They just used what it had and adapted to the information it was getting from the world.
DANIEL: And that works for language too. It’s a bunch of strategies that we use to do perceptual stuff in the real world.
MARK: Exactly. I think that basically what I’m saying is that our neural hardware is very plastic and responsive to the environment, but this means that we make use of that developmentally, in that we don’t build as much internally inside. We don’t have to carry that much information in our genes. We can just have enough to make the plasticity and then the baby pops out and does the learning to finish off.
DANIEL: And that makes sense because language changes way faster than our genetics change, so.
MARK: Exactly. I mean, we’re learning that our gene and gene expression changes faster than we thought but still, we’re also learning that language changes faster than we thought.
DANIEL: That’s right. Rebecca Shapiro — I think it’s okay if somebody’s been a guest on the show and they answered this question, I’ll just refer to them by name — Rebecca Shapiro, (hi, Rebecca) says, this was on Bluesky: rules that are learned and then broken. Isn’t that an interesting sort of thing?
MARK: I like that. Yeah. And if I can leap in here, I think there’s something really crucial in this. I have a paper that’s come out in the last few months with Uta Reinöhl looking at the metaphor and particular properties of syntactic constructions around metaphor.
DANIEL: You’ve done a lot with metaphor.
MARK: We’ve done a lot with metaphor, and we have this paper in Protolang about proto language from a couple of years ago in the Journal of Primatology.
DANIEL: We talked about that on the show. And if anybody has forgotten, do you mind if I just recapitulate?
MARK: No, please.
DANIEL: So, you noticed that A: no other animal except humans seem to use anything like metaphor.
MARK: Yeah.
DANIEL: And also, metaphor may have been a way that we got syntax, because if you use expressions more or less literally, you can start dropping things off. So, you can say, “Oh, I drove 5 miles to the house, and then finally I arrived.”
MARK: Yes.
DANIEL: Or I can say, “I arrived at the house.”
MARK: Yes.
DANIEL: I can drop that off if I want to.
MARK: Or, “I arrived there,” if you want to be a bit shorter. But you can just say, “I arrived.”
DANIEL: But if I pick up ARRIVE, and I start using it metaphorically, like “arriving at a conclusion”, I can’t drop anything out, it sounds weird. “Tell me your conclusions and how you arrived.”
MARK: [LAUGHS] By bus?
DANIEL: You have to say something like “at them.”
MARK: You do.
DANIEL: Even something like “have to,” right?
MARK: Yeah.
DANIEL: You can HAVE a thing. But then we started talking about HAVING obligations metaphorically. And I can’t start dropping words out like, “Do you have a baseball?” “I have.” “Do you have to go?” “I have.” You’ve got to include “to,” right? “I have to.” Even something as seemingly vacuous as that, you’ve got to include it. And because you have to include more words when you’re using things metaphorically as we started using things metaphorically, that means we had to use set phrases, which means set word orders, which implies grammar and syntax.
MARK: We would say that with the metaphors, as you say, you have to include words that you don’t have this motivation. You’re not saying them because they’re not in the context. You’re just having to express them to support the metaphorical interpretation, which means then that their position, remember we talked about information structure, which says that how informative something is guides where it is in the sentence. These words aren’t informative about the context because they’re just there for semantic needs of the metaphor. You would leave them out if they didn’t have to be expressed. And so, that means they don’t have this guiding information structure to tell them where to go in the sentence.
So, to reduce our own pain as interpreters or speakers, we give them a conventionalised order. And that then, we would argue, generalises to all, from the metaphorical cases to the rest of language. And so, we start building this structure of rigid orders and grammatical structures, which can be about order or about endings in different ways. We build that around the metaphorical constructions and extend that to literal.
DANIEL: So, that’s the work.
MARK: That’s the work.
DANIEL: Now, take me from there. Now that we understand it, where do we go from there?
MARK: What was the question were dealing with?
DANIEL: It was rules that are learned and then broken.
MARK: Yes, exactly. So, I think that’s a really nice case. We can think of metaphorical constructions, the type we’re talking about, as breaking rules. They’re basically, when I talk, for example, about arriving, I’m implying that where I’m arriving is a physical location. It’s like you don’t have to have me tell you that because we know arriving has a physical location.
DANIEL: And used to be specifically, to a shore, adrīpāre.
MARK: Really? I did not know that. That’s so cool.
DANIEL: If you arrived someplace, it had to be a shore, but then we extended.
MARK: It had to be the shore.
DANIEL: I might have said this last time we talked about this example.
MARK: Oh, really?
DANIEL: I don’t know. I’ll go through transcripts and find out.
MARK: Yeah. That is so cool. So, yes, we widen our scope, but also sometimes we just break it. And we break it because we can do cool things when we break it. If I put together two words, let’s say, “I ate my computer,” then you think, “Okay, he probably doesn’t mean a literal…” It doesn’t make sense. The computer is not a thing you can eat, so therefore, you’re working hard to try and think out, okay, what interpretation could there be by relaxing the semantic constraints on what these words can mean, you try and figure out a figurative interpretation of this combination.
DANIEL: Ah, when you eat something, it’s no longer available. So, maybe he means that he destroyed his computer?
MARK: Yeah, exactly. By using it too much for things that were helpful, but then it got destroyed, for example, like could mean that.
DANIEL: Yep.
MARK: And so, I think this is a really nice example of we have these rules in how we use particular words and then the creativity of languages, putting them together and breaking those rules.
DANIEL: I love it. That’s so good. That explains a lot. I like this a bunch. Elizabeth on Bluesky says: change is the only constant. That’s an interesting l… Language change is very huge.
MARK: Yeah, it is. It is. I was pondering possible answers to the challenge you gave me, and I was thinking one way we could do it is by describing change. And whatever language we have is just whatever you can get out of the processes of change. And we’ve already touched on lots of that, and we’re going to talk about change a lot. Change is inherent in language, but it does speed up and slow down. Writing seems to have slowed down change in some ways for us. I think that’s really interesting.
DANIEL: James on Discord says: language is a collaboration. This came up in an episode with Ruanni Tupas, how you can’t really abnegate or it’s not right to abnegate your responsibility to make conversation happen even if the other person doesn’t speak the same way you do or something. It’s tempting to chuck it and say, “Oh, this conversation didn’t work because this person’s no good at English.”
MARK: Oh, my goodness, you miss out on so much though if you do that.
DANIEL: Yeah.
MARK: I got on a train once in… Where was I? I was in Berlin, and I needed to get to Poznan in Poland, and my fiancé at the time told me, this Russian train, and nobody will know about it. And you get on, you have to actually haggle with the conductor, then you buy a ticket, and…
DANIEL: It was a terrible idea.
MARK: It was fine. I got up to the platform, and this secret that nobody knows about, there’s hordes of Australians waiting to get on this train…
DANIEL: Oh, really, okay.
MARK: …in Berlin to go east, and I end up in this sleeping compartment with a guy who only spoke Russian and German, and I had basically no German and a few words of Polish at this stage, in terms of Slavic.
DANIEL: Okay.
MARK: We worked out how to speak to each other. I got to understand. I learned that he lived in Russia. He was ethnically German, and that gave him the right to work in Germany. He was coming to Germany and working, taking money back to Russia. We learned a lot about each other, even though we had very little language in common. But you can find a way if you want to. And I think we miss out so much if we give up on what’s not easy.
DANIEL: Adam Schembri on Twitter says — it’s one word and an emoji: Usage! and then face with tears of joy. Which made me think, “Hmm, what did that mean in the instance in which he tweeted that?” Which is a very usage-based question, right?
MARK: Exactly right. Yeah.
DANIEL: Usage, a thing is what people mean when they use it.
MARK: I think this is very much a picture from that kind of a functionalist view of language. But I think there’s definitely a fundamental truth there that we cannot divorce language from its use. It’s not a thing we can study in the abstract and ignore its use. There’s not an ideal.
DANIEL: Well, what do you mean? There are entire schools of linguistics that ignore usage and focus on an ideal speaker here. [LAUGHS]
MARK: It’s true.
DANIEL: It’s true. And you will learn certain things from that, and you will miss a lot of other things.
MARK: Yes.
DANIEL: Luisa had one.
MARK: Luisa had one.
DANIEL: So, Luisa Miceli, you’ve been hanging with Luisa.
MARK: Exactly. Our colleague and collaborator of mine, University of Western Australia. And Louisa, I asked her what she thought the fundamental idea or generalising concept from language was, and she said this really great thing. We kind of came to this way of expressing it as wasted potential.
DANIEL: [LAUGHS] Oh, oh, oh, okay. I know what you mean, but I want to hear it from you.
MARK: So, it was lovely, because Luisa has been thinking about listening to lectures by Chomsky and thinking about the Chomskyan program for linguistics, which is, she says she has quite a positive spin on her, which is surprising, given a lot of Australian linguists are kind of not so positive about the generative program. But her idea is that Chomsky was asking a different question to what we normally ask here, which is more usage based. She’s saying, he’s trying to find out what’s the limits to the creativity that’s available within a language. And when you ask, what’s grammatically possible in a language, you’re kind of asking, what’s the limits of our creativity in terms of putting things together?
DANIEL: What’s possible?
MARK: Exactly. What’s possible in terms of what’s the kind of creative limits we can use. Maybe we can break even those rules when we have a really good reason. But usually in this program, they try and think, can we think of any context where we want to break these rules? And they try and study the rules that we are not very likely to want to break.
And Luisa was saying that one of the things she found fascinating about language is that the boundaries of what we can say as possible sentences are really huge. It’s like this is a vast space, whereas the kind of usage, what we actually say today is really super predictable often and it’s…
DANIEL: Wasted potential.
MARK: Wasted… We came up with this term, wasted potential. She doesn’t mean that as quite as negatively as it sounds, but it’s like…
DANIEL: That’s good though. That’s right, because Hedwig points this out. We can say sentences because there are literally infinite number of sentences, but we usually just say, “Can I have the toast?” Or, “That’s my flannel.”
MARK: Exactly.
[LAUGHTER]
DANIEL: “How’s it going?” “Not bad.” It’s an 80/20 thing or maybe even a 95/5 thing.
MARK: But I don’t see this as being such a big problem, because I think that what’s going on is that, and this is very natural when we look at the generative program, it’s about what’s possible and not possible. Whereas for some of us, and I think, for example, Adam, it’s like, what do we actually say frequently? Like, it’s the distribution that matters for many of us. When we look at the distribution of what we say, it’s a very relatively narrow distribution compared to the space of what once in a million years, we might say potentially. And that difference is really important in the way you look at language. If you look at it as a distributional thing, then you’re interested in what are we likely to say and how do our expectations impact what comes out. And that’s in contrast to ignoring what’s likely to be said and just trying to put some hard boundaries on the edge.
And I think I find it’s harder to believe in those boundaries because, yes, we might have taken care of the easy context, the likely context, and said, “Okay, we wouldn’t say this in those likely contexts.” Are we sure we’ve tested every possible context to know that we could never say this sequence of words as a sentence in English?
DANIEL: Yep.
MARK: Maybe we could.
DANIEL: All right, well, thank you, everybody, for all of the suggestions that you’ve made. And I think there’s a lot to chew on here, and we could stop the episode right now and have some good stuff. But I’ve got my answer. And you’ve got your answer.
MARK: Yes.
DANIEL: Let’s get some more coffee.
MARK: Let’s get some more coffee. And let’s do answers that matter.
DANIEL: Answers that matter. Oh, the other ones.
MARK: The other ones. I realise I shouldn’t have said that.
DANIEL: Now, I’m going to give my answer.
MARK: Yay.
DANIEL: And then, I want to hear your answer. And once again, the question is, what one thing explains the most about language? And I posed this question to my son because I was giving him a ride to the airport this morning, and I said, “That’s what this show is about. What one thing explains the most about language?” And he said, “Oh, is it that people are stupid?” And I said, “Oh, you guessed the thing.” Because for me, the one thing that explains most about language is language takes the form it does because of human cognitive limitations.
MARK: That sounds really cool. I like it.
DANIEL: Okay. And I had three other answers, but I realised that they were but tributaries to the river. And the river is language takes the form it does because of human cognitive limitations. So, humans have limited cognitive power.
MARK: Yes.
DANIEL: And to overcome that, we use little hacks.
MARK: Yes. That’s so true.
DANIEL: And so, we can study those hacks.
MARK: Yes.
DANIEL: A lot of the language things we’ve been talking about are these hacks. So, I want to talk about the hacks. Putting things into words is slow, but decoding the words is fast. So, we have to take shortcuts. Also, we have limitations on working memory, so we have to keep it tight. We get tired or distracted, and we don’t want to pay close attention all the time. And there’s noise in the channel.
MARK: Yes, that’s so true.
DANIEL: Okay, so here are the hacks. I thought of these in terms of speaker-side hacks and hearer-side hacks.
MARK: I love that.
DANIEL: One is implication.
MARK: Yes.
DANIEL: So, encoding things is slow, but decoding things is fast. So, if I want to keep your attention and you’re not bored all the time, I’m going to have to do certain things to speed it up. I’m going to have to make my words mean more than they say. And that’s where you get implication.
MARK: Cool. I like that. That’s really interesting.
DANIEL: So once again, if you say simple things, you mean simple things, but if you say complicated things, you mean complicated things. Hedvig’s example was, “She closed the door. She caused the door to close.” If I say, “She closed the door,” that means she closed the door. But if she says, “She caused the door to close,” that’s a little more complicated, so it means something more complicated than she closed the door. What did she mean? Well, I don’t know, but whatever she did mean, it wasn’t she closed the door. Because if she closed the door, I would have said she closed the door. I meant something else, so I said something else.
MARK: Yeah, this is interesting, but it depends. Going back to, remember we were talking about the inflection from morning and afternoon and night in Gamilaraay?
DANIEL: Yep.
MARK: We see this in languages. Some of them inflect for a causative structure on the verb. So, they can say have a verb like, “The door closed,” and then they put an ending on the word for close to mean to make it then a causative where I close the door and it’s their normal word, like, “I closed the door, and, “I caused the door to close,” through this suffix, they just mean the same thing, but in…
DANIEL: Is it more complicated?
MARK: …English, it’s not the same.
DANIEL: Hmm, okay.
MARK: Because we have two ways of saying it, and if you choose the more complicated way of saying it, then why are you choosing that more complicated way of saying it?
DANIEL: We used to have a causative in English.
MARK: Well, we have a zero causative…
DANIEL: Yeah.
MARK: …because “the door closed” and “I closed the door”.
DANIEL: Yeah.
MARK: We just can’t see that’s eroded with time. But there was a suffix that was a causative on verbs in Germanic.
DANIEL: Okay, but they can do complicated things even language with causatives.
MARK: They can do as complicated as we do.
DANIEL: Yeah, okay.
MARK: It’s just like we talked about. They will just use a different mechanism, and their mechanism, maybe it’ll be more complicated than our mechanism to get the same distinction, or maybe it will be less complicated.
DANIEL: Yeah, okay. But anyway, implication is one way that we use shortcuts to overcome a bit of a processing lag that would bore our listener otherwise.
MARK: I would agree with that. There’s something interesting though that comes up there. You said uncomplicated expressions should correspond to uncomplicated ideas and complicated ones to complicated ideas. But is an implication having an uncomplicated expression corresponding to complicated ideas? Isn’t that what implications are there for?
DANIEL: In that case, maybe I should say, if you say something in an expected way, you get an expected…
MARK: I think that’s the difference. It’s like you prop it up with your expectations that it would mean this isn’t such an uncomplicated expression. It’s more complicated than we need, given how much we already know that then pushes us to take implications.
DANIEL: Okay, okay. Next thing that this explains, and this is a speaker-side hack, ambiguity.
MARK: Yes.
DANIEL: I can use a word that is more broad or more general.
MARK: Yes.
DANIEL: If I want to say something, I could pick a word that’s close, like thing. It was a whole thing. And now what I make you do is I make you do the work of processing which version of thing I mean.
MARK: And that work is not a problem if it’s primed enough.
DANIEL: We’re good at it.
MARK: We’re really good at it. And we have evidence that given the expectation about the thing that’s going to be talked next… Well, I’ve forgotten the technical, there’s a technical term for this, will come back to me in a second. But when what we do is we use less precise language. The more expected that item is, so in other words, the clearer the idea the listener has about what’s coming up, the more vaguely we will express ourselves. And that’s because using vaguer language is easier. When we talk about minimising speaker effort, vaguer is easier. It seems, as far as we can tell, that’s like something that develops an ease, that develops because of the dynamics or the distributional properties of language. Things that are frequent are easy. Things that are vague are easy.
DANIEL: Mm-hmm. Mm-hmm. My partner and I, we know each other so well. We share so much context that we can say things that we don’t want the children to understand in such impossibly vague terms they will never catch on.
MARK: Exactly. I’ve seen episodes of spy dramas where spies have worked together for years, can talk to each other while their enemies are overhearing, and still communicate without betraying their secrets.
DANIEL: Very good. Okay, next one, metaphor. We’re kind of dumb. We don’t understand new stuff well. So, any way that we can take old knowledge and transplant it into new knowledge is going to help us.
MARK: Yeah, that’s the thing.
DANIEL: That’s what metaphor does.
MARK: That’s what metaphor does. And we would say this, myself and my colleague, Uta, who noticed first this thing about the necessary expression of all components of the metaphor, I think we would see this as kind of fundamental to how language is developed. It’s the source of creativity within language, and that’s both in terms of creating new language structures over time. But also, in the moment, expressing something new that we don’t have words for. We just put together pieces we have, and we push that square peg through the round hole, make it work.
DANIEL: [LAUGHS] And it works. Okay. Yeah, metaphor is just so crucial and so pervasive. We use it just all the time. And then, the last speaker side hack that I have is common ground. We have a knowledge of each other’s context, or we know that we don’t know each other’s context. So, we’re able to draw on mutually accepted knowledge. And that affects language because if you know something, if a piece of information is mutually known, we don’t have to say it. And if it can be guessed from context, we don’t have to say it.
MARK: That’s right.
DANIEL: So now, we get elision or ellipsis. So far, I’ve got implication, ambiguity, metaphor, and common ground. Those are huge.
MARK: Common ground. Do you mean, like, what we’ve got so far? What’s been communicated so far?
DANIEL: The pile of information between us that we both mutually accept as so.
MARK: This is something really interesting in trying to work out how to express the model of language that includes things like common ground, one of the things I realised is we can bring in lots of examples from technology, because these hacks that are good for language, they’re good for every type of communication, which is why we use them in technology. So, the example I used… I don’t know if we talked about this last time I was here, the way MPEG encodes video.
DANIEL: You did mention that.
MARK: Ah, I’ve said everything already.
DANIEL: I edited this recently. Yeah. So, the MPEG compression standard works because it doesn’t encode every single pixel on every single frame. If it doesn’t change from one frame to the next, it just leaves it and says, “Only give me the changes.”
MARK: Give me the changes and it…
DANIEL: And then, it gives a keyframe every once in a while…
MARK: Exactly.
DANIEL: …to reset.
MARK: Yes, and it’s such a good idea that we do it too.
DANIEL: Yeah. [LAUGHS] If it doesn’t change, don’t change it.
MARK: You read a book and it’s like using pronouns inside the chapter or inside the paragraphs, and you get to the end of the chapter, and then you start mentioning everybody by their full names or descriptions, and then you start using pronouns again. And it’s just like we reset. At the end of the chapter, we kind of do a partial reset over our context, and it’s like we get a full picture frame again and then we work on incremental changes to that.
DANIEL: Yep. It’s a compression algorithm.
MARK: It’s a compression algorithm.
DANIEL: Okay. Now, I’m going to switch to the listener-side hacks, and these are things that I, the speaker, can do to help my listener. Everything so far has been making it easier on me, but now I’m going to make it easier on you.
MARK: This is what we often call audience design.
DANIEL: Oh, okay.
MARK: So, it’s basically what the presenter, the speaker is doing to make their communication more accessible to their audience.
DANIEL: Okay. Okay. First one, agreement. I make things agree. For example, I don’t say two book, I say two books. The “s” is a plural morpheme. Little redundant because I already said two, which implies plural, but I’m going to do that, and what that does is locks it in. If you missed it the first time, if you had a lapse of attention, you don’t miss out on the whole thing. If you miss out on the whole thing and cause a breakdown, then I have to repeat the whole thing again, which is very inefficient. So, agreement helps to make up for lapses in listener attention.
MARK: That’s true. Although what this… I can see this doing like an error detection job, but it doesn’t do an error correction job. Like, if I miss the two…
DANIEL: Okay.
MARK: …and I hear books. Sure, I know it’s plural or think it’s plural, but I don’t know how many. So, it’s not like it gets me the two back again.
DANIEL: Okay.
MARK: But it might help me, like, “I should have paid more attention.”
DANIEL: Maybe I’ll try to gather that up in the next thing that he says, and then I’ll ask a clarification question if I still miss it.
MARK: But it’s interesting that this hack isn’t something all languages do. So, one of the languages I’m learning is…
DANIEL: Not all languages use agreement?
MARK: Not of this kind.
DANIEL: Okay.
MARK: So, learning Persian at the moment, and I’m sorry if I get this wrong, my knowledge is imperfect, but my understanding is you don’t use the plural marker. They have a plural marker you use on every occasion that works for every word. There are other plurals, like conventional plurals, but that you don’t use that plural marking if you use a number.
DANIEL: Interesting.
MARK: But you happily use it like if you just want to indicate a plural when you don’t. I’m not sure. It’s a good question, is it obligatory? I don’t know. So many things I need to learn.
DANIEL: So many things to learn.
MARK: But the point is you don’t use it if you use a number. And so, some languages make this kind of choice.
DANIEL: What kind of agreement does Persian use? Do you have to… Like, subject-verb agreement or is there anything?
MARK: Subject-verb agreement, but really, that’s it. That’s all I can think of. There’s no grammatical gender. Even the pronouns are like genderless. They’re male, female, you don’t specify. In the pronouns, and I can’t think of any other. Obviously, when you talk of multiple events, there’s kind of timing agreement, which is a little bit different. So, the way the tenses work when you’re coordinating, talking about different time combinations is a little bit different in Persian than in English. And so, you get people, I think of using the conditional when there is Persian speaking English. But it’s a frequentative, I guess, is what it’s corresponding to. But yeah, they have much less agreement. And I think that’s an impact of contact with Turkish.
DANIEL: Yeah. Okay. I was also thinking, gee, it’s wild that there’s an Indo-European language that ditched gender on pronouns.
MARK: Well, it’s…
DANIEL: It’s a contact…
MARK: Hindi doesn’t have it on pronouns either…
DANIEL: Oh, okay.
MARK: …but it has it on the verb. Like, it will mark it on the verb, but not on the pronoun.
DANIEL: Right. Okay.
MARK: And nouns are very gendered, but in Persian, they don’t, they got rid of all of that.
DANIEL: Wow. Okay.
MARK: [LAUGHS] That’s so cool.
DANIEL: Next one. Redundancy.
MARK: Yeah.
DANIEL: And this is something I do to help you, because once again, if you missed it the first time, then I can make other words jump in. A lot of people really complain about redundancy. “Why do we say it’s a FREE GIFT? It’s redundant. It’s not efficient.” Well, actually, it’s okay for words to jump in and help out.
MARK: It is so okay. And it also changes the meaning. We were just discussing this the other night where it’s like you put two things, they look like they’re redundant, but actually changes the meaning of the combination.
DANIEL: Yeah. They’re not…
MARK: A bit like a metaphor, so they’re not really redundant. So, if you’re an advertiser and you’re saying FREE GIFT, what you’re trying to say is, this is a gift without strings attached. But if an advertiser says it’s hard to believe, but that’s the intent they’re trying to say. And it’s the same… like an example of we talked about is, “There was a mouse on my desk,” and you show no surprise. And so, it was the mouse-mouse.
DANIEL: Oh, a mouse-mouse. Oh, all right. Oh, my gosh, I was expecting a mouse, but I didn’t expect a mouse-mouse.
MARK: Exactly. And so, this is a case where we get to the literal meaning of the word. If we double it up and it sounds like, “Oh, you’re being super redundant. You’re saying MOUSE twice,” but you’re not really. You’re changing what you’re referring to.
DANIEL: Mm. And once again, that helps to make up for any lapses in attention or I think it helps to disambiguate as well.
MARK: Real redundancy would… Yeah, this is disambiguation, but I don’t think this is a real case of redundancy because you’re changing the meaning, you have to hear both mouses to understand this restricted meaning.
DANIEL: That’s right. So, it’s just a case of semantics being underspecified once again. Usually, I can get away with under specified semantics, but when I really need it, I can pull it out.
MARK: You can pull it out. Exactly. Yeah.
DANIEL: Next one. Dependency. We have lapses in working memory. It only stretches so far. So, we tend to keep, for example, adjectives and nouns very close together. You’ll see RED HOUSE or HOUSE RED or something like that. But you won’t see HOUSE and then the rest of the sentence and then RED. It’s got to stay within phrasal boundaries, sort of. And that just makes sure that we haven’t forgotten what the things pertain to. They don’t stretch out because we have limitations on working memory.
MARK: I think that’s a really good point. I think there’s a lot of syntax we can predict if we say related words, as in semantically connected words, should be close to each other. I think that gets a lot of the degrees of freedom out of word order in languages. I mean, occasionally you get that people being poetic, and Sydney University’s motto. I might have mentioned this in the past as well, but it’s “Sidere mens eadem mutato” which literally means “same mind under another star”, but STAR is the first word, another is the last word CHANGED, literally. MUTATO is the last final word. But then, they put SAME MIND together in the middle. Yeah, mens eadem.
DANIEL: So, it can happen.
MARK: It can happen. But it’s usually, once again, if you’re doing the kind of slightly weird thing to some effect you want to generate. And so, what they’re doing here is they’re building the expectation with the star, and then they go to the same mind, but then they give a bit more punch to the changed at the end, putting mutato at the end means that it’s a really different place. They’re really different stars.
DANIEL: Yeah, okay.
MARK: And so, I think that if we play like language is always letting us play with these things of building up expectations about what’s normal, and then we break that. Something’s going on.
DANIEL: Yep. Something’s going on. Finally, the last listener side hack is prominence, which we’ve already talked about. I make something more prominent, and that directs your attention. So now, what I’m doing is I’m manipulating your attentional focus.
MARK: Yeah.
DANIEL: Because you might not know what to focus on, I’m helping you.
MARK: Exactly. And that’s really good when you think of that as helping the listener understand the balance that they should be using or the direction they should go between what they’re expecting to hear and what they need to listen to. And it’s painful if someone’s always using this stuff, because we can’t maintain that much attention for that long, because most of us, like you’re saying, we have limitations on our cognitive processing. So, it’s really helpful if people selectively point out the bits that are actually worth paying attention to.
DANIEL: It helps.
MARK: It helps.
DANIEL: So, that’s what I got. So, once again, just to summarise, language takes the form it does because of human cognitive limitations, and those limitations are working memory, attentional focus, processing speed, which is related to attentional focus and noise in the channel. So, what have I explained? I explained implication, ambiguity, metaphor, and common ground. Agreement, redundancy, dependency and prominence.
MARK: Wow.
DANIEL: There’s probably more. So, this explains a lot.
MARK: It does.
DANIEL: Okay, so that’s what I got.
MARK: Well done. I like it.
DANIEL: How about you?
MARK: I think what we’re going to find is that what I have as my fact is overlaps so much.
DANIEL: Okay, cool.
MARK: With everything we’ve talked about from our Discord listeners to even what Luisa suggested to what you’ve suggested. So, I would say language is for expressing what’s new.
DANIEL: Language is for expressing what’s new, okay.
MARK: So, what this means here is that when we are creating language instances like words, sentences, discourses, we are wanting to lead our listeners, recipients into new information or new situations, and we want to actually focus on what’s new. And focus is… Like you said, there’s various tactics we use to actually guide the people that we’re talking to what’s new. And it breaks down very much like what you described in terms of listener side things as well as speaker side things. So, the speaker, in trying to express what’s new, is going to minimise expressing what’s old.
DANIEL: Because if it’s already known and accepted, you don’t have to say it.
MARK: You don’t have to say much.
DANIEL: Much. You need to say less.
MARK: You need to say less.
DANIEL: Okay.
MARK: Because the person we’re speaking to might not know what we’re going to talk about next, but they might know there’s a small bundle of things they’re likely to talk about next, but they might not know which one we’re going to pick up to continue with. So, we might need to say, like, use a pronoun or use a small definite phrase to say, this is what we’re going to pick up to continue the presentation of new things with. But the point is, like you’re talking about listeners getting bored, we don’t want to bore our listeners. And so, we try and squeeze down the information we communicate in order to express the stuff that’s not new but make enough of a launching pad to express what’s new.
Yeah. So, I think that this hides underneath it. It’s like an iceberg. When we say expressing what’s new, it’s like we’re talking about the tip of the iceberg, what’s underneath the surface comes from what does it mean to not be new? And that’s a real question. So, we’ve already talked about implication and like how if we’re talking about… Let’s say we’re talking about a car, then the car’s door, car’s roof, car’s tires, they’re not new anymore because we mentioned the car. So, these implied non-newness.
DANIEL: A thing can be really complex, composed of multiple elements, but by giving it a name, we refer to all that stuff.
MARK: We refer to all that stuff. But also, even if we are just referring to the car, let’s say I’m saying Daniel’s got this great new car, people will automatically assume that there’s tires involved. Like, tires will not be unexpected now, if I start talking about tires. If I say it’s got beautiful black tires, that won’t be so unexpected. It won’t be a big surprising thing. I won’t have to introduce it so much. But if I start talking, let’s say before electric cars were common and they talked about, “It’s got a huge battery,” I would have to say more because people go like, “Do you mean is that for charging people’s phones or is that…?”
DANIEL: “Is that the battery, the lead acid battery that starts the car?”
MARK: Exactly.
DANIEL: “No, no. It’s not that. Let me tell you more about it.”
MARK: Exactly. We have to expressed more because it’s so new. And so, the newer something is, the more language time we have to actually devote to it. And we want to devote to it because that’s what we’re wanting to express, is what we want to focus on as the exciting thing for our listeners is the new stuff. And it’s the fun stuff to talk about. So, there’s this interaction between what’s predictable, which is the key thing in terms of what’s not new, defining what’s not new, then launching off what’s predictable into what is new.
Let me take a little diversion into one type of not newness that I don’t think we’ve talked about so far. And that is to do with the topographic structure of language.
DANIEL: Topographic structure of language?
MARK: Yes. And this is not about map making.
DANIEL: Or it might be.
MARK: We didn’t really know about topographic maps, which show the shape of the countryside. But here, what we’re talking about is topographic mappings, as in neural mappings in the brain are often called topographic, where they connect nearby inputs, as in nearby in terms of either their actual location in the retina or so on, or their function to nearby bits of the cortex, physically nearby. And in general terms, we can think of this as similar inputs mapping onto similar outputs.
DANIEL: Okay, so catch me up here. Are we talking about where words are mapped in the brain?
MARK: We’re not going directly there. Let’s start at the abstract. Using the abstract idea, we’re talking about how similar representations in language correspond to similar things in the world.
DANIEL: Okay, so for example.
MARK: So, for example, let’s think about numbers. We have a similarity relationship between numbers. Like, 5 is closer to 6 and 4 than it is to 2 or 9. And then, if we look at, for example, temperatures, 5 degrees is closer to 4 or 6 degrees in terms of temperature than it is to 9 or 2 degrees.
DANIEL: So, there’s a topographic relationship between the numerals and the size of the number of things…
MARK: Conceptual numbers and temperature. Exactly. And we find this with so many things like maps. If you look at your street map, it’s like things that are close together on the map, close together on…
DANIEL: Close together in real life. Okay. Okay, good.
MARK: Yeah. That’s probably the best example.
DANIEL: Okay, good. So, let’s see. Once again, the concept is we express what’s new. We don’t express what’s old. Or if we do, we express it in a different way or more compactly.
MARK: Yeah.
DANIEL: Okay.
MARK: So, how do we know when something’s old? We know it’s something’s old if we’ve already said it before.
DANIEL: Yeah, I’ve already heard this. Good.
MARK: Or somebody’s describing the same situation even if they’re using new words, if we worked out they’re describing exactly the same event.
DANIEL: Oh, yeah. You could express something old with brand new words, but I would know it.
MARK: Exactly. We’ll work that out and go, “I’ve heard this story before.”
DANIEL: How do I do that even if you’re using totally different words, how did it work out that it’s old stuff you’re talking about?
MARK: Exactly. Because we have two sides to the communication. We’ve got the representation side, and we’ve got what’s being represented. And so, as we’re building up the model of what’s being represented in the current version of the tale, you go, “Oh, that sounds very familiar.”
DANIEL: Okay.
MARK: “This is just Rapunzel again.”
DANIEL: Yeah, yeah, yeah.
MARK: And I’ve heard that story before, albeit in different synonyms.
DANIEL: Okay.
MARK: But in terms of where this topographic mapping story comes in is it tells us about what’s new and some things, new is a gradient thing, it’s not an absolute. So, it’s not just it’s completely new or it’s completely old. It’s like some stuff is pretty much like what I’ve heard before, and we’re always trying to squeeze away the bit that’s already present, that’s already known.
DANIEL: Let me use an example. Let’s say that I’ve got… I want to say John had a sandwich…
MARK: Yes.
DANIEL: …and so did Maria. I could say, “John had a sandwich and Maria had a sandwich,” but that would be a weird way to do it.
MARK: Yeah.
DANIEL: So, I would say, “John had a sandwich and so did Maria.” Now, how do I know that referred to the same kind of event when I used different words?
MARK: Yeah. It’s because we have this kind of ability with words to refer back to other whole references. So, in this case, it’s referring back to the bit that’s not the subject. In this case, having a sandwich is what the “so” refers back to. And we picked that up again using this like pronoun for verbs.
DANIEL: Yes. It’s like using pronouns for verbs.
MARK: It’s a pronoun for verbs.
DANIEL: Okay.
MARK: And they’re really useful things. All these pronouns, ways of avoiding repetition are so useful.
DANIEL: Okay.
MARK: We want them all the time.
DANIEL: Okay, cool.
MARK: Yeah.
DANIEL: All right. So, you’ve shown me that a very big thing about language is expressing what’s new and avoiding what’s old, at least in the same exact way.
MARK: Yeah. I think there’s so many things about this that… So, one thing that’s just occurring to me now, for example, is when we’re talking about the… We talk about the arbitrary meanings of monomorphemic words?
DANIEL: Words with one morpheme in them.
MARK: Exactly. So, one meaning unit in them.
DANIEL: Yep.
MARK: And they have this kind of arbitrary relationship between the thing and the meaning.
DANIEL: Yep.
MARK: Is a standard belief in linguistics, and that’s more or less true.
DANIEL: Sandwich, if we called it a schnoof, it would work just as well.
MARK: It would work just as well.
DANIEL: Yep.
MARK: So, yeah. The arbitrariness of this relationship, what’s really cool is that even within the category, like all sorts of different things we can call sandwich, and that’s great. And variation, the way we say sandwich mostly doesn’t matter very much. And this is part of this kind of arbitrariness within the category is a way we’re reducing the amount of newness we have to pay attention to. So, it’s not newness itself that’s a problem in terms of processing effort and what we want to actually have in our communication. It’s newness that matters. So, every time I say any of these words in this sentence, I’m saying them differently to how I’ve said them before, but it’s not a newness that matters.
And it’s the same thing in terms of the meanings we have in mind when I say sandwich, the newness of the sandwich. If it matters, I’ll talk about it if it’s a newness that matters. If it was not a newness that matters, I won’t talk about it more.
DANIEL: And then, my brain says, “He’s describing this sandwich in a way that he hasn’t done before. Maybe there’s something relevant about that.”
MARK: Exactly.
DANIEL: Okay.
MARK: So really, when I said language is for stating what’s new, I should say language is for stating what’s new that matters.
DANIEL: Yes.
MARK: And we try not to dwell on the newness that doesn’t matter. There’s a lot of it. There’s always newness. Like that person who said everything changes in language, it’s true. But we’ve got to get good at ignoring the changes that don’t matter. That’s part of what language does.
DANIEL: Yep.
MARK: And paying attention to the changes that do matter. And we see that in so many different levels of language, building phonemes out of sounds, that’s ignoring a lot of changes to focus on a bunch that will say, these ones matter, and every other difference between sounds won’t matter.
DANIEL: Like, if I say [t], [t], [t] three times, those are all different.
MARK: They’re all different.
DANIEL: That doesn’t matter.
MARK: Exactly. And if you’ve ever learned a foreign language that has different aspiration patterns in English than English does, you know it’s hard work.
DANIEL: Yeah. I mean, in English, it doesn’t matter. But suddenly, I’m in a context where it does matter, and I’ve got to figure out how to make it matter.
MARK: Yes.
DANIEL: Or not matter.
MARK: But what’s really cool is our brains are really good at this stuff. Like, I was learning Scottish Gaelic, and there’s palatalisation of consonants, some of them are really huge, and you can’t miss them because it’s difference between “ta” and “ja”. But some of them, it’s like “ta” and “pya”, you’ve got to pay attention. And it took me a long time. Then one day, it’s like I woke up in the morning, it’s like, how could I not have heard that difference before? It’s so obvious. And that was really useful because I then learned Polish where it also matters, and it was there already, and that made life so much easier.
DANIEL: The sun has come out.
MARK: Yes. It’s a difference that matters.
DANIEL: It’s a difference. And it seems relevant because I think we might be glowing with a kind of halo from the knowledge that we have imparted. Okay, well, I feel like we’ve explained a lot about language with a very small set of facts.
MARK: I think we have. I think there is definitely a lot of language that’s covered from this relationship between the old and the new.
DANIEL: I think maybe we should express our bias. There’s something that I’ve noticed about myself, and I mentioned this earlier. The only kind of linguistics that I think is interesting anymore is cognitive linguistics. [LAUGHTER] That’s the only kind I care about now. If it’s cognitive, I’m like, “Ooh, that’s interesting.”
MARK: Yeah.
DANIEL: And I feel like we’ve expressed a lot of things from the cognitive point of view and not so much from sociolinguistics or other kinds of…
MARK: That’s a very good point. I think that what you’re stating is true. And I have to say, let’s say I was a computational linguist primarily with historical bent, and then I went to a cognitive science department, and I got cognitivised there, and by the end of my time in Edinburgh, I was very much interested in how people process language as the kind of the fundamental like… maybe it’s reductionist, but it’s the idea that what’s interesting in language is what either impacts the way we process it or is an outcome of how we process it.
DANIEL: Because that influences how language looks and operates.
MARK: Yeah, but I think you’re right. We cannot look at this necessarily in terms of individual processing. And this is why I’ve been excited being involved with work, like with Nick Fay here at UWA in the past, and looking at doing experiments on people, not individually, but in groups, and seeing how people create communication systems when they’re given various restrictions. Like, you have to communicate about these things, and you have this medium in which you have to communicate, and you have no conventions, you’re not allowed to use conventional stuff.
DANIEL: They invent conventions.
MARK: People are amazing. For example, we get people to describe tangram figures, which are these kind of abstract figures made of little triangles, and we say, “You’ve got to communicate to your partner which of this list of tangram figures you’re talking about at any moment, that you have an ordering, and they’d have a different ordering, but they have to identify which one as you go through them in your ordering, they have to identify.” And at the start of the tutorial session, where we have people do this, and we say, “Okay, you can describe them in language,” but people at the start are making lots of eye contact, and they’re using lots and lots of words to describe these figures, “Oh, it looks like an ice skater, but they’re falling over to their left,” and blah, blah. And then at the end, by 30 minutes later, it’s like they don’t look at each other. They just say one word, and they’ve got it.
DANIEL: Because they added that to the common ground. And this happens in groups. Groups is where it happens. Not individually.
MARK: Exactly.
DANIEL: Because language is a collaboration.
MARK: Well, not everybody agrees with us, Daniel.
DANIEL: I know, I know, some people. I was looking at James’s answer, language is a collaboration, but some people really do feel like language is internal.
MARK: For me, when I think about language being internal, I kind of think about… I can pretend to be two people, and then I can language inside myself. But that’s kind of a part of the process, is I am having half of a dialogue, which is a monologue, or I’m actually trying to have both sides of the dialogue inside myself.
DANIEL: Douglas Adams once wrote, “The problem with talking to yourself is that half the time, you already know what you’re going to say,” and then Arthur thinks, “Only half the time?” [MARK LAUGHS] Well, Mark, thank you so much for hanging out with me and explaining language. And this has been a lot of fun.
MARK: It’s been a lot of fun. Thanks, Daniel. I really appreciate it.
DANIEL: And thanks once again to our patrons, listeners, and friends on social media for giving your answers and contributing the things that you do, it’s great to have you. Great to have you be with us on this journey through language and figuring stuff out.
MARK: Thank you.
[Transcript provided by SpeechDocs Podcast Transcription]