In what was meant to be a casual chat, cognitive scientist Dr Mark Ellison answers galaxy-brain-level questions about how language works.
- Why aren’t we more efficient with language?
- How do we know when something has gone wrong in a conversation?
- Why don’t we just talk in a flat monotone all the time?
- Why do fairy tales start a certain way?
- Why is it so tiring to speak another language?
Fortunately, he helps us keep our eyes on the ball for this episode.
Listen or watch this episode
Video
Audio
Patreon supporters
Huge thanks to all our great patrons! Your support means a lot to us. Special thanks to:
- Iztin
- Termy
- Elías
- Matt
- Whitney
- Helen
- Jack
- PharaohKatt
- Lord Mortis
- gramaryen
- Larry
- Kristofer
- Andy B
- James
- Nigel
- Meredith
- Kate
- Nasrin
- Joanna — bumping it up to the Supporter level after years of being a patron!
- Ayesha
- Moe
- Steele
- Margareth
- Manú
- Rodger
- Rhian
- Colleen
- Ignacio
- Sonic Snejhog
- Kevin
- Jeff
- Andy from Logophilius
- Stan
- Kathy
- Rach
- Cheyenne
- Felicity S
- Amir
- Canny Archer
- O Tim
- Alyssa
- Chris W
- Felicity G
Become a Patreon supporter yourself and get access to bonus episodes and more!
Become a Patron!Show notes
The Loom of Language by Frederick Bodmer
https://archive.org/details/in.ernet.dli.2015.283471
Indogermanisches etymologisches Wörterbuch by Julius Pokorny
https://archive.org/details/indogermanisches02pokouoft
Nicholas Fay: The Interactive Evolution of Human Communication Systems
https://onlinelibrary.wiley.com/doi/full/10.1111/j.1551-6709.2009.01090.x
Paleo-Hebrew Alphabet
https://www.paleohebrewdictionary.org/alphabet/
The Ankh | World History Encyclopedia
https://www.worldhistory.org/Ankh/
Ankh | Ancient Egypt Online
https://ancientegyptonline.co.uk/ankh/
George Miller: The magical number seven, plus or minus two: Some limits on our capacity for processing information.
https://psycnet.apa.org/doiLanding?doi=10.1037%2Fh0043158
George Miller’s Magical Number of Immediate Memory in Retrospect: Observations on the Faltering Progression of Science
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4486516/
Human speech may have a universal transmission rate: 39 bits per second
https://www.science.org/content/article/human-speech-may-have-universal-transmission-rate-39-bits-second
Transcript
[Transcript provided by SpeechDocs Podcast Transcription]
[Because Language theme]
Daniel: Hello, and welcome to Because Language. A show about linguistic science of language. My name is Daniel Midgley, and this is going to be a slightly different episode. You see, my friend, Dr Mark Ellison, we featured his work on the show a few times. He came to Perth recently, and he stayed at my place for a visit. We thought why don’t we have a chat about language for the show and talk about what he’s been doing in the way of research?
Well, what started as a casual chat led to a galaxy-brain-level explosion in which I understood way more about language than I did when we started. Admittedly, this is the thing that happens when you talk to Mark. So, what we’re going to be talking about here in this chat is what we are doing cognitively when we’re using language in a conversation. Here are some concepts we’re going to talk about.
Prominence. Why we make things stand out in a conversation, and what that means. We’re going to talk about attention. We’ve paid just about enough attention to make sure we understand, and that has an influence on how language looks and how it works. We’re going to talk about expectation. We have a sense of what we expect to happen, but when we don’t get what we expect, we try to find out why. Sometimes, that means something’s gone wrong, and we need to fix it. Or sometimes, it means there’s a little extra meaning that our interlocutor wants us to read between the lines. There’s also manipulation. We use things like intonation and pitch and volume and silence to guide the interpretation of our listener, even in ways we’re not aware of. These things help us to understand and remember what the other person is saying.
All of these things that I’ve mentioned and all these things we’re going to talk about have implications for how language works in a situation. And really, they have a lot of impact on how language is the way it is. So, here we are. Dr Mark Ellison.
I’m here with Dr Mark Ellison of the… what’s your affiliation?
Mark: My affiliation is the Department of General Linguistics [unintelligible [00:02:24] at the University of Cologne, but my employment fund comes through a very large collaborative research center funded by the [unintelligible [00:02:38], the German Research Council. This collaborative research center is investigating prominence in language.
Daniel: Prominence.
Mark: Mm-hmm. Prominence, exactly.
Daniel: Some things are prominent and some things not so much.
Mark: That is pretty much the definition of prominence is this.
[laughter]
Mark: You have a bunch of things that are otherwise on the same level or from the same category or in the same situation, and some of them stand out and some of them don’t. And that’s prominence.
Daniel: Some of them you mark, and some of them you don’t.
Mark: That’s another way of putting it. Yeah. Depends on whether you look at it as a thing that has just the way the world is or something somebody did, is whether it’s that way or someone did it for a reason. And both are interesting.
Daniel: Now, we’ve been hanging out this week, we’ve been talking about language…
Mark: We have.
Daniel: …a lot and other stuff. And we’ve mentioned your work on the show before a bunch of times. But we’re also pals.
Mark: We are, very much.
Daniel: So, it’s been really cool to get to hang out like we did in the Perth days. That’s nice.
Mark: Those were pretty remarkable days back in 2010, particularly.
Daniel: Now, we’re here in the interview spot.
Mark: In the interviewee couch.
Daniel: This is apparently the Because Language interview area and we’ve got our coffee.
Mark: We have.
Daniel: We’re at the start. You started out in maths?
Mark: I did. Kind of at university, my first degree was in pure mathematics. But I would say if you scrape back the history a little bit further, I went to university with the ambition of becoming a theoretical physicist because I got really into Einstein and general relativity equations and things like that while I was still at school. It was really fun, and I wanted that but things went a little bit wrong with that plan.
Daniel: Okay. As they do.
Mark: As they do. I went to physics class, and I got really bored. [chuckles] It wasn’t the fault of physics as a whole. It was the fault of having linear algebra thrown into this by pile of mathematics, applied mathematics, and physics in that order, all in first year, and I was just bored out of my mind with it.
Daniel: Okay, so you made a lateral move to maths.
Mark: Actually, I focused more on computer science at first, so I had a general practice through my years. It’s like something I discovered, [unintelligible 00:05:23] my practice at the end, which was the subjects that I got bored in, I generally did worse in as the year went on, and then at the end of that year, I would drop them.
Daniel: Okay.
Mark: At the end of first year, I dropped physics and I dropped philosophy. Once again, not because I’m against the general idea of philosophy…
Daniel: I know you’re not.
Mark: …but the implementation at the university, I felt that there were lots of very specious things being discussed for hours on end for no purpose in philosophy. So, I dropped that as well. Math split into two parts, pure mathematics and applied mathematics. I have both of those and computer science as my three subjects for second year.
Daniel: Okay.
Mark: In third year, I dropped computer science.
Daniel: You were finding your way.
Mark: I was finding my way through a process of elimination of the weakest links. [laughs]
Daniel: It’s subtractive. It really is. A lot of things are subtractive.
Mark: A lot of really good things are subtractive. I really encourage people to do more subtractive thinking.
Daniel: How did you bump into linguistics and what made you stick with that?
Mark: That’s a great question. I started, I would say, at about the age of eight when I discovered that my father could read Greek and I thought, “That’s pretty cool.” He lent me or passed on to me his grammar of Greek. So, I learned the alphabet first, and then I learned…
Daniel: Let’s see, eight years old, this is when you were in your second year of university?
Mark: [laughs] Life would have been slightly more interesting if I was, but sadly, I was still in a primary school, which was tedious in some ways, but probably good for my socialisation.
Daniel: Yeah. Unskippable training missions.
Mark: Exactly.
Daniel: What was it about Greek? Was it just the interest of cultures of antiquity or a vibe, an aesthetic that you imagined with Greek?
Mark: It’s a great question to ask that one because I was also… since it was the only foreign language I had encountered at that stage really in any kind of quantity, through my father would come with us to church and instead of his English Bible, he would have the Greek New Testament with him.
Daniel: That is so like early Christian.
Mark: [chuckles] In church, when people would have discussions about what a verse means…
Daniel: [chuckles] Well, actually…
Mark: …he is like, “Yes, the Greek says this.”
Daniel: Whip it out.
Mark: How can anybody trump that?
Daniel: No, we can’t.
Mark: I think I enjoyed that aspect of it, but it was the only one I encountered. But then, I discovered books in libraries, books in bookstores. I think I was 10 when I bought three Teachers of Language books. One was Arabic, one was Chinese, and the other was Bengali.
Daniel: I’ve done that and then I never followed through.
Mark: I obviously did not learn any of those languages from those books. I did, however, read a lot of them, as in a lot of the books without memorising it all, sadly. But I did learn a lot. For example, I learned about Bengali script without even conquering that in its completeness, but I understood how it worked from that. At the same time, in encyclopedias, I’d read about Sanskrit. I was very fascinated with India even at that same age, eight, nine, and studied a lot from encyclopedias about that. And then, I found a page with a description Sanskrit alphabet, which is one of the very few things I organised to get photocopied for me when I was a kid and carried around for years.
Daniel: At a time when photocopying meant something.
Mark: It did mean a lot. That was the kind of situation that kicked it off. I just kept collecting language, books about languages. I also read from the city library, The Loom of Language, I think. Was it Bloomfield? Bloomfield or something. And I love that book.
Daniel: How old were you when you were reading Bloomfield?
Mark: 9 or 10. I used to borrow my brother’s library card because he was old enough to be reading, have a ticket for the adult section of the library.
Daniel: Oh, wow.
Mark: I would borrow [chuckles] his card. There’s a funny story about that. One time, I came down to borrow… clearly, like, I’m not 14, which was the age limit for that library, and I’m handing the card to the person with the books, and they said, “Oh, yes, this card is expired, and you need to get it renewed.” And I just ran away.
[laughter]
Daniel: “Hey, wait a minute. He’s trying to check out aspect on the theory of language.
Mark: That’s right. [chuckles] I’m not 100% sure of the author, but the book was called The Loom of Language, and it was a beautiful book that talked about historical linguistics and Indo-European and the relationships between Indo-European languages. Also, things about how languages work in the present.
Daniel: Okay, this is me as well, because when I was a kid, I used to go to the library at the elementary school, library in the school where I first received my tutelage. I was in the 400 section reading Comfortable Words by Bergen Evans, who was a personality in the USA, and had a radio show about language, and he was full of phrases like, “Why do we say these phrases?” It’s just so weird that as young people, we got ourselves… look, we gravitated toward these things that we would eventually still love…
Mark: Take over our lives. [laughs]
Daniel: …40 years later. Isn’t that funny?
Mark: It is funny, it was interesting…
Daniel: People don’t change.
Mark: People don’t change. I think there’s a lot of stuff we’re born with. I probably am into that even more than you are, Daniel, but I think there’s a lot of stuff we’re born with that sticks with us through our life. For me, this obsession with language just continued. Whenever I had money, I would buy another Teach Yourself Language book. But I had no chance to learn a foreign language. I had two years of mediocre French instruction at high school, which had no support. And then after two years, the school got rid of the whole thing and there’s no languages taught at my school after that. So, there wasn’t any chance to do anything linguistic-y until when I went to university. I wasn’t focusing on the language side of things in my studies because I had these things I was really into already like maths, computer science, philosophy, and physics, and I want to do languages as well, but there just wasn’t room in the program. So, I wasn’t very good at socialising. I was very shy first year, and I would spend my lunchtimes in the library looking at books and discovering things, like Pokorny’s Indogermanische Wörterbuch and Manisch Wörterbuch. This was basically in the 400 sections of the research library at Sydney University, [chuckles] instead of socialising with my peers [crosstalk] the first year.
Daniel: One of the things we noticed…
[Kookaburras cackling]
Daniel: Couple of kookaburras.
Mark: Giggle chicken.
Daniel: The birds always go a bit nuts here.
Mark: They do.
Daniel: Language covers a lot of things, but what are you into? Knowing your work, I want to say, how it all began, but there’s a strong part about what’s going on in our brains when we’re doing language. How would you describe the area of language where you live and work?
Mark: Wow. That’s actually not an easy question for me either, because I keep getting my horizons broadened that I will start being involved with something because of my background in, say, maths and stats can help a project, but I’ll find the project really boring. Then, once I get into it, I find, “Oh, this is really interesting and important, and I wish I’d thought about this before.” [chuckles] So, I would say at the moment, if I have to summarise, that I’m interested in the four-dimensional structure of language, particularly as it’s embedded in or relates to cognitive processing. Okay, so this means cognitive processing and the origins of language is one way of dealing with that one boundary in the four-dimensional space of language.
Daniel: Okay. Four-dimensional space?
Mark: Meaning space and time.
Daniel: Okay, got it.
Mark: Okay. Mostly we’re two-dimensional on the surface of the Earth but hey, some people live in high places. [chuckles] And I’m interested in what happens right now when we use language.
Daniel: Okay. Right now, we are sitting in a place and we’re using language, we’re talking to each other. If I had to make a statement about what we’re doing, there’s a lot of things going on. We’re both saying stuff, we’re communicating, we’re getting ideas across. To do that, we have to put the ideas in our minds into words in a comprehensible way. While that’s going on, we’re listening to each other, we’re making predictions about what the other person is going to say. Sometimes, there are surprises, but we do a pretty good job of it. Over time, more and more information gets added to a kind of pile that we both accept as being the case. We do it for a certain purpose, which is to have fun or to inform or to put out an episode of Because Language.
Mark: Exactly. I think we have purposes, and those are really crucial for defining the space of this shared information structure that we build together. I think one of the kind of insights, I think that is important in understanding conversation is this movement away from a model of, “I am sending information to you,” which is the super simple, but a great starting point as a model of communication. But in interactive communication, we need a richer model, which is that there is this shared structure that we build up together. And I can work out, for example, as while I’m speaking, given this structure, what do I need to say in order to extend it with the things understandings that I want to add to it. Part of that is that I can also… given that you have this structure as well, I can say, “I don’t need to say so much about this thing here in order to add the new understanding, because Daniel will be able to predict from what’s in the model already.” A lot of what I would say, and I just need to channel the discussion in with a few words in a particular direction, and Daniel’s brain will fill in the rest [crosstalk] say.
Daniel: That’s when it becomes complicated doing linguistic communication, because you’re not just communicating to me. We’re also communicating to an unknown audience who doesn’t share our understanding. On one level, yes, we two of us are making guesses about each other’s knowledge and beliefs, desires, intentions, plans, and goals. And then, we both know what’s on the stack. We both know what’s on the pile of stuff. Then, when you want to add a pile, you have to say, “Okay, Daniel doesn’t know this. Let’s add that to the pile of common ground.”
Mark: We do. That’s exactly how it works. We add to the common ground what can’t be predicted. That’s what we communicate, what can’t be predicted. But in communicating that, we’re leading the person to bring in their own predictions about what’s happening to fill in gaps. As you say, we are talking for the audience as well as for each other. So, we are going to add a bit more explicitness in that construction because there is a wider variety of backgrounds out there. There’s some really fun work that I did with Nick Fay from the University of Western Australia here, and that work was based on things he had done at the University of Glasgow with Simon Garrod amongst others. This was communicating using something like Pictionary. So, it’s called the Pictionary Task.
I would have a list of 16 concepts, and then you had a list of those 16 concepts as well, but in a randomised order. What I would do is I would start drawing the first concept on my list, and you would have to work out which one it was. As soon as you worked out which of your 16 items it was, you would say, “I’ve got it,” or communicate that through a button press or something. And then, we move on to the next drawing. But you could also interact in the drawing process. Now, what’s interesting and relevant for our situation here is that they did studies where there’s two people communicating, just dyads, and they play together with the same 16 concepts. Many, many games, many, many run throughs that list of 16 concepts.
Daniel: Over time, you get better and better, so you need to say less and less.
Mark: Exactly. That’s what happens. The amount you have to draw gets less and less. The other person says, “Oh, I know that corner he’s just started to draw. That’s the corner of a museum, so he’s saying museum,” or something like that. However, what’s really interesting is the comparison between what happens when two people play 42 games of this together versus when in a group of eight people, everybody plays with each other person in the group, six games each. They play six games with one person, then they swap partners, six games with another person until everybody in the group has played with everybody in the group. What we find is that they both reduce the amount that they draw in order to communicate about the same amount. So, everybody gets equally concise. But there is a real difference. What’s preserved in the shortened drawings is different. The difference is that if you show a person who’s not been involved in the whole drawing process, the little drawings you have at the end, the kind of micro drawings you have at the end…
Daniel: Won’t make any sense.
Mark: …won’t make any sense, except the ones that the individual pairs did will make less sense than the ones that were created within the group structure. So, those preserve more of the iconic nature of the signs because they’ve had to deal with new people.
Daniel: It’s got to look more like a museum.
Mark: Exactly. Even though it’s still very abstracted away, it’s got to look more like a museum. So, in the groups, one of the fantastic things people did in terms of finding a great representation with a museum was like half of a bone. They draw a stylised end of a bone, and they only get halfway through drawing the whole bone and then pressing it, “I know what that is.” [Daniel chuckles] Someone, if they look at their list of 16 concepts, not many of which involve bones, even if they’ve never been participating in that sequence of games, they’ll look at that and go, “Hmm, that’s probably a museum.” Whereas if they look at what the individual person did, which might have been the corner of a three-dimensional room where there was a vase sitting in a box, and just that corner now becomes the stylised representation of museum, three lines joining the point, I don’t know what that means.
Daniel: The consequence of talking to multiple people that you don’t know as well.
Mark: Exactly. There are real consequences for that, like good consequences in terms of the way a language works. This relates to what we’re doing because though we’re doing this dyadic interaction, we have to try and pretend we’re in a big group so that our communications are actually able to be more understandable for people who don’t have the same level of shared background that we have.
Daniel: It’s fascinating that we’re making guesses about each other’s knowledge, but I can pretty much guess for you. But I’m also having to guess the knowledge of multiple unknown people who are very different from each other.
Mark: Yeah. And we do this really well. However, one of the things that I think is really relevant to this is how many successful podcasts have either an interview format or discussion format. That’s because interaction with two people, you do a lot better than just one person on their own. And that’s something we also have a Pictionary study about, but we don’t have to talk about that now.
Daniel: Okay. That’ll be good for later.
Mark: Yeah.
Daniel: One of the things that happened recently was that I was doing an interview for Adelaide and the ABC, the Australian Broadcasting Corporation, like I do every week, and they wanted to talk about redundant expressions. They were expecting me to say something like, “Oh, what about silly expressions like ‘free gift’? Why do we say ‘free gift’? Why can’t we be more efficient? Why can’t we just say ‘gift’?” “What about an ‘unexpected surprise’? Why do we say that sometimes? Why don’t we just say a ‘surprise’? It’s more efficient. Why do we have redundancy in language?” I came to you and I said, “Do you have a bit that you do on redundancy?” And you did.
Mark: The more I think about it, the more so much of my work relates to redundancy. I was just realising the paper that we published this year in the International Journal of Primatology, I’m really excited about that.
Daniel: We talked about that in the show, yeah.
Mark: Did you? So good. In this paper, one of the things we talk about is this debate in the origins of language, did language start off being something holistic, where one utterance that you couldn’t break down in any way meant a whole bundle of things, like…
Daniel: What is a tiger?
Mark: Exactly. Where there’s no separate bit of it that meant tiger and no separate bit meant run. No separate bit meant danger. No separate bit meant, “Look up, get ready to do something dramatic.” They’re all just one lump. The alternative view is the view that there were all these separate pieces. There was like tiger piece, and you wouldn’t know what reaction was wanted, like, “Cute little tiger come playing in the distance.”
Daniel: “Look, I’m pretending to be a tiger.”
Mark: Exactly. It could be any of those things. The idea was that you had these isolated meanings and we learned to put them together. With a holistic view, the idea was we had these complex meanings and then gradually, some of them accidentally had bits in common that we then reanalyzed as being part of us having an identifiable, separate meaning. We suggest a third alternative, which relates to something that we find present in all languages these days, which is selectional restrictions.
Daniel: Selectional restrictions.
Mark: Yeah. Let me give you an example. If we talk about arriving somewhere, then we tend to think that the thing that’s doing the arriving is a physical object. An object with physical location properties. The destination, the ‘at’ component of the arriving that also has some properties, like being a physical location.
Daniel: I arrived at the house.
Mark: Exactly. You don’t think, for example, that the ‘at’ argument is going to be an emotional state, for example, or an abstract concept like addition. However, we can use these things metaphorically, but that’s not the point yet. We’ll get to that. We can talk about that later. But the point now is that when we say arrive, we have this focus meaning of like transitioning from one place to another and arriving at that place. However, there are these extra restrictions it’s putting on what type of thing can be involved in the process and the context in which that happens. It’s like the word kind of spills over. The meaning of the word spills over to the things that are around it when you make up sentences. The subject of who’s doing the arriving and the destination ‘at’ argument about where the arriving, its destination, properties of those are forced on it by this word ‘arrive’.
Daniel: They walk in together.
Mark: Exactly. That’s a great image. They all have to fit together. Otherwise, other things, weird things happen. We can talk about those. The key thing here is all the pieces interlock because they’re overlapping, and our sense of words overlaps.
Daniel: And that’s where redundancy comes in.
Mark: That’s a wonderful example of redundancy because when I say ‘arrived at the hotel’, for example, ‘arrived’ says my argument for destination has to be a physical location. One of the properties of hotels is they are physical locations. So, there’s redundancy there.
Daniel: If you say, “Daniel arrived at the hotel,” Daniel works… there are things that Daniel fits in to arrive because Daniel is an agent of being who is human and has human ideas…
Mark: And is able to have a different changing location in space. Exactly. That overlap in the senses and the meanings is without that, we get in a horrible jumble. We try and rescue ourselves. We do our best to make sense when things are mismatching, but sometimes we can just completely fail, which can be used for comedy purposes and so on. If we can make sense of it, even though it’s a fail, then we call those metaphors.
Daniel: Yeah, okay. Redundancy is not a bad thing in language.
Mark: For example, we think that redundancy was key to getting language going. We think that redundancy, that ability to have things interlock and then have things mismatch a little bit, led to an explosion in the kind of meanings we could create through metaphor. For example, when I say, “Daniel arrived at a conclusion.”
Daniel: That’s a bit of a mismatch, a little tiny bit.
Mark: It is a mismatch. If we get this feeling, “Oh, it makes sense because it’s a process.” And then at the end of it, you’re somewhere where you weren’t before. It’s like you have this sense of there’s a completion of a process.
Daniel: Don’t forget that the word ‘arrive’ itself is the result of a mismatch of a language change, because when the word was first coined, it had to be a river, ‘adripa’.
Mark: Really? I did not know that. [laughs] That is hilarious.
Daniel: When you arrived at a place, it had to be near a river.
Mark: Wow. That is so good.
Daniel: Gradually, we changed it to expand it. We could apply it to more things. The selectional restrictions widened a little bit and now we can arrive at conclusions.
Mark: We see this happening again and again. This process of building metaphors, which you can just widen the application, but you can also have this kind of metaphorical shift happening. This is something my colleague, Uta Reinöhl, studied in her PhD work, was this widening when the word for ‘middle’ in Sanskrit eventually became a post position, just meaning ‘in’ in Hindi.
Daniel: Okay, so you would take a part of the word middle and you’d stick it after another word.
Mark: You put the whole of the word middle and you combine it with the noun as like the possessor, the middle of the river, for example, if we’re talking about a river. And then now in Hindi, the descendant of that word mádhya, in Sanskrit has become má in Hindi. If you put má with a noun like river, it may just means ‘in the river’.
Daniel: Wow.
Mark: There’s an interesting shift there that’s really important for the way language has developed. We see this as where if you can kick off language by having these messy, overlapping semantic units that you put together and they have to match up. What happens is when you start to break things, making metaphors, we found that there’s a requirement that both bits of the metaphor, both of the bits that cause the break have to be present, have to be said explicitly, even in languages where you can leave out anything.
Daniel: We talked about this with your paper. You could say, “I started walking across the road, but I stopped in the middle.” I don’t have to say, “in the middle of the road.”
Mark: Exactly.
Daniel: But if I would describe myself as a middle-of-the-road kind of guy, if I take that and expand it metaphorically, I can’t just say, “I’m a bit of a middle.”
Mark: “I’m a middle kind of guy.”
Daniel: No, you have to say, “I’m in the middle-of-the-road kind of guy.” So, your work found that when it’s more literal, you can start omitting things. When you pick those expressions up and use them metaphorically, you can’t omit so much. You have to be more specific.
Mark: We would say it slightly differently because you use this kind of more middle more you have to say more explicitly. As far as we know, it’s a categorical thing. There’s so few things in language that so many things we thought were categorical and they’re not. This thing just seems to be categorical
Daniel: You mean like there are things that are literal and then there are things that are metaphorical?
Mark: We think that one is harder, but what we would say is there seem to be things that are literal… when things are literal, people can leave out lots of stuff. When things are metaphorical, you absolutely can’t. It’s not like some things are more metaphorical, and so you can, there’s a gray area. We haven’t found a gray area yet.
Daniel: “Daniel walked all day and then he arrived,” that’s fine.
Mark: Yeah.
Daniel: But if I say, “Tell me your conclusions and how you arrived.”
Mark: You can’t.
Daniel: You can’t. You have to say how you arrived at them.
Mark: It doesn’t matter how much… normally, we say or expect, that if you can have a zero argument, all you need to use it is you need to have the thing highly primed, which is technical speak for you need to have already talked about that destination recently. We can have it as close as you like. Like you just did in the same sentence, you had the destination named, but you still have to say “at them.”
Daniel: Even something like, “I’m doing this because I have to.” “Have” to is kind of vacuous. It’s omitting things. But I can’t say, “I’m doing this because I have.” Even something minimal like to that is so seemingly vacuous. It’s like the beginning, it’s not even a full infinitive. It’s like an infinitive minus its verb. You still have to do that if you’re not literally having it. If you’re saying it’s a kind of obligation.
Mark: Exactly. I think this… I’m not exactly sure of the historical processes for this construction.
Daniel: Oh, I am.
Mark: Are you? Tell me.
Daniel: Yeah. This is one path that probably is the most likely candidate. You say, “I have a box,” or no, “I have a letter. Here it is. It’s in my hands. I possess it.” Then, you say, “I have a letter to write.” Now, it’s in the future a little bit. Then, it swaps, “I have to write a letter.” Finally, you can drop the letter entirely and say, “I have to write.”
Mark: Yes, that makes sense.
Daniel: That’s the path that I’ve seen proposed for that.
Mark: That makes sense because, for example, in German, you still see a lot more of the “I have a letter to write” construction.
Daniel: Ah, interesting.
Mark: Ich muss einen brief schreiben.
Daniel: With German propensity for verb last.
Mark: Yes, exactly.
Daniel: Okay. Language starts out kind of literal and moves to metaphorical.
Mark: Let’s say that it actually jumps to metaphorical.
Daniel: Okay.
Mark: Remember this idea that words have these extra requirements they put around them about other words that they go with. As soon as you have those requirements, you can break them. When you break them just that kind of forces you to like, if they didn’t break them, it’s easy to understand. It’s really straightforward. Like, “I’m arriving at Daniel’s house,” if I break them a little bit but still can make sense of it, I’m arriving at a conclusion. Then it’s pushing us to think about arriving differently in this context. You can do that as soon as you have these selectional restrictions, that these fuzzy boundaries of words, which is actually, we think the natural why would have words have clear, sharp semantic boundaries? It’s unlikely. It seems really unlikely to us.
Daniel: The next part of that is because people started using language metaphorically and had to include certain things, then you start getting regular patterns where things have to be said and in the same way. So now, people are building up verbal habits.
Mark: Exactly.
Daniel: And that builds up into grammar. What else about prominence have you noticed? Because you’ve done work recently…
Mark: Lots of different kinds of prominence.
Daniel: [laughs] Okay.
Mark: I have done two different kinds of prominences that I’ve been really focused on. One is about how we refer to things. Do we use pronouns or names, definite descriptions like the black cat or do you use Felix or do you use ‘it’, or ‘he’ or ‘she’ depending. That’s one type of work. The other type of work I’ve been doing is with people in phonetics, looking at what’s the impact of different intonation on remembering things. That’s really fun.
Daniel: Let’s talk about intonation. We were talking earlier about an example where, “I think Mary took the book.” “No, I don’t think Mary took the book. I think John took the book.” I say that little… what even is intonation, what even is stress? I think stress on John. “I think John took the book.”
Mark: Yeah. There’s a few things that happen, one thing that we see a lot, and it’s really the focus on what we do in Cologne, it doesn’t mean it’s the only thing in the world, it’s just what we’re focusing on, is intonation, which means how does the pitch that somebody uses in their speech change in response to what they’re trying to do with their speech? Ask questions, focus your attention on different bits of the sentence. They might also be varying the loudness.
Daniel: So, I said, “I think John took the book.” I said it a little louder.
Mark: There is also, if you listen very carefully, the pitch goes up.
Daniel: “John took the book.”
Mark: Exactly. Partly, just because it’s longer, you lengthened it, gives more time for the pitch to arrive. We have these things we call pitch accents, which often go up and down. These kinds of things are often used. They are used in many languages to mark something out as being this is a little bit… something you should pay a bit more attention to. What we have been doing to try and get at this in its purest form is to look at what is the impact on people trying to remember a sequence of nine digits without any repeats. Basically, the digits one to nine just scrambled in order. In this case, the experiment has been done in Italian and in German. At the moment, we’re looking at reporting on the German study. People are given the sequence of digits through headphones, they hear them. And they’re actually produced by recordings of each of the digits being said with different intonation patterns. And then, they’re combined electronically so we’re sure that the actual intonation patterns are exactly the same.
Daniel: Same every time.
Mark: Same every time. Just rearranged according to what we want to do as experimenters.
Daniel: I might hear 7-3-4-2-6-1-9-4-3.
Mark: Exactly. That was a really nice example because you gave what we call the boundary fall intonation to these. So, they’re grouped…
Daniel: Three numbers.
Mark: Exactly. We don’t have so much in terms of the spacing between the groups, but what I noticed in the way you said it was there’s a fall at the end of the last word in each of those examples.
Daniel: 9-4-3.
Mark: 9-4-3 and so on. This was one of the conditions in the experiment. We did this…
Daniel: Oh, you mean not everybody heard that?
Mark: No, we had four different conditions. Everybody got exposed to all four conditions. And because we didn’t want to be dealing with individual differences and abilities to remember sequences of the numbers, we wanted to actually look for what’s the effect of the different conditions. If everybody sees all the conditions, that helps us tease apart what’s different between individuals from what’s different between the conditions [crosstalk]. Here, conditions means, sometimes we do it with falling intonation at the end.
Daniel: 9-4-3.
Mark: Others, we had 9-4-3 with no falling intonation. Just the nine was pronounced the same way as the three, just flat. Then, we had two other conditions that were really fun. One was put a big emphasis on the last word in the triple, so 9-4-3.
Daniel: 9-4-3.
Mark: Exactly. The emphasis was, in this case, the pitch accent, I’ve probably made it louder and longer, which is not so good, but it should be 9-4-3 with a rising pitch on the third one. We had that at the end, the last one in each group of three, except the very last one always had a falling off intonation to mark the end of the sequence so people knew when they got to the end. The end of the first group of three and the second group of three had this strong accent on the last word. That was one of the interesting conditions. The other one was instead of a falling pitch at the end, it was 9-4-3, 2-6-7 [crosstalk] rising intonation at the end.
Daniel: Okay. You were looking at whether they could remember, whether that raised the prominence enough for them to remember what the number was.
Mark: Not number, but the whole sequence. We were interested in their accuracy across the whole sequence, but also individual positions, because if you do better on all the individual positions, you do better overall.
Daniel: That’s true.
Mark: What we found was really surprising, we think.
Daniel: Okay. Can I predict?
Mark: Go for it.
Daniel: Okay. I predict that when you use flat intonation, they’re not going to remember it.
Mark: So true.
Daniel: Okay. When they use rising intonation, probably not going to remember it that well. They might remember the next thing.
Mark: Okay.
Daniel: And then, when you use bumpy intonation, probably better.
Mark: Bumpy intonation mean that stressing that third one in the sequence.
Daniel: Yeah. Okay. So, what did you really find?
Mark: What we really found was you’re exactly right. If everything’s monotone, people do really badly. It’s shocking how badly they do it.
Daniel: There’s nothing to grip on to.
Mark: Exactly. There’s so much space in the graph. In all conditions, they start out with the first guy. They can remember that really well. The last one, mostly pretty well. It’s just in between we see a lot of difference. The fall off as you go through that first position, second position, the third position, and so on, you get a fall in accuracy in all the conditions. But in the monotone case, it’s a straightforward dive to inaccuracy. People do really badly when they get to the sixth item on the list, they’re just like, “Is it 6? Is it the third one or the fifth one? I don’t know. One of those number things.” That was the worst, as you predicted, and exactly right intonation. Intonation of some sort really matters. If you don’t have intonation, everything’s flat, people are doing really badly.
The next worst one of our four was this falling intonation at the end. It’s a grouping, but it’s little fall at the end. 2-3-7-6-4-9, blah, blah. That was interesting because we were surprised by that because for many of us it’s default way you group things.
Daniel: That would be why because it’s the default. There’s nothing special about it.
Mark: That is very potentially the case. In a way that we’re talking about just earlier, if you have a default pattern, then that’s telling you when something is interesting, it’s only interesting if it departs from that default pattern. If you follow the default patterns, you’re telling your listener, “Nothing interesting is happening here. Don’t pay so much attention.”
[laughter]
Daniel: This is so weird that we’re so sensitive to this.
Mark: We are so sensitive, it’s shocking. This is the thing. I didn’t know about this and I wasn’t interested in prominence before getting involved in. Now, it’s like the most interesting thing in language. It’s like telling us where to look.
Daniel: But now, the best-remembered sequences?
Mark: Okay, so we’ve got two left. There’s one where we actually accent the word. There’s one we have a rising boundary at the edge of the third word in this.
Daniel: 9-4-3 vs 9-4-3.
Mark: Exactly, 9-4-3. What we find is that if you accent the third word in the group of three, you get better accuracy on that word, but not the others. It improves accuracy very much on that word. That word is getting all the attention, but people are no better with the other words.
Daniel: If I say 9-4-3, don’t remember the three.
Mark: Yeah.
Daniel: [chuckles] Okay.
Mark: But if you say 9-4-3, then they remember the whole group better.
Daniel: Wow.
Mark: Which completely shocked us. The first time we did this experiment, I came onto the project later, but my colleagues were a bit surprised.
Daniel: Well, but we do do that, don’t we? We say, “You guys, I am so stressed out. I’ve got to write an email, I’ve got to wash the dishes, I’ve got to hang out those clothes.”
Mark: We do do these things. We know all the tricks. Psychologists and researchers like myself, we’re finding the tricks by little bit but humans use this all the time.
Daniel: It actually helps us to remember things better.
Mark: It helps us to communicate better because when we’re telling somebody, like in that listing that you just gave, that’s because you want your audience to pay attention to all these tedious things you have to do if that’s what you are expressing and you want them to see each one because it’s tedious and you want them to feel you’re paying…
[laughter]
Daniel: When I said, “I don’t think Mary took the book. I think [stressed] John took the book.”
Mark: Exactly. So, they have a false belief. They’re making a prediction. They’re thinking, you’re going to say Mary took the book. You’re doing your best to disabuse them of that false prediction.
Daniel: It takes a lot of difference, a lot of prominence, a lot of effort to counteract their wrong belief.
Mark: Exactly. We are prediction engines. One of the things that’s really come out in recent thinking about this that’s built on a lot of experiments is that what we do, the way we cope with getting so much information thrown at us all the time is we make predictions, and we just pay attention to the difference between our predictions and what actually happens. One of the really lovely examples of this, which is appropriate for a lot of our audience, I think, is if you’re ever watching the cricket, like in the summer that we’re coming to, there will be cricket on TV. If you watch a fielder out there catching a ball, one of the instructions that you’ll hear the commentators at lunchtime, when they give instructions to little kids about how to play cricket well, it’s like, “When you’re catching the ball, keep your eyes on the ball.” The reason for that is because when you’re looking at the ball, you’re continually making predictions about what should the ball look like if I’m in the right position to catch it. If you watch the cricketers catching the ball, they’re staring at the ball, and then they’re adjusting. They either running one side or the other side, running back more, forwards more in response to how they’re seeing the ball. It’s because they’re making a prediction about how the ball should look if they’re in the right position to catch it, and then adjusting their movement so that when the ball finally reaches close to their hands, at a level of their hands, they’ll be in exactly the right place to catch it.
Daniel: And that’s what we’re doing when we’re conversing. I’m making guesses about what you’re going to say, and if I get things that I expect…
Mark: Then, I don’t pay that much more attention.
Daniel: Everything’s fine. I can minimise the amount of effort.
Mark: Exactly.
Daniel: But then, I get something I expect. You say something extra loud, or you say a different word that has weird selectional restrictions, it doesn’t fit in with another word that you said before.
Mark: It attracts attention.
Daniel: Then I’ve got to figure out what you meant by that. A little earlier, I said something that wasn’t right, and you had a model of what I knew and what I was going to say, and then I said something really, really different, so different that it was wrong. And then, you realise, “Oh, Daniel misunderstood. I’ve got to repair this.”
Mark: Exactly. It was just like that. That’s how it happened. It’s interesting you mentioned that when things don’t match up as well, because I think that’s one of the big reasons why we use metaphors, which is exactly where things don’t match up, is we do that because it drags in a bit more attention. Poets love metaphors because that draws the reader’s attention into the words…
Daniel: …so to speak.
Mark: So to speak.
Daniel: And then, you remember it better. We remember things better because they’re different from what we’re expecting.
Mark: Exactly. One of the real joys in this is that through this mechanism, you can find really wonderful ways of expressing things that they somehow give you a little bit of joy every time you see them because they are so good. Let’s take an example back from this Pictionary experiment I described. When people saw a bone for museum, they loved it. You see one person thinks a bone in this community of eight people, and as we have all their drawings, you can see it. When somebody encounters someone who’s using bone for museum, they go, “Wow, that’s really clever.” And after that, they adopt it. They use that immediately and forget whatever they were doing before, [chuckles] except for one person who refuses to use bones. Come on, clearly everybody else knows it’s better. Why can’t you just get on board?
Daniel: I think that we’re seeing that right now. Right now, it’s November 2022. We’re starting to get into Word of the Year season. Everyone’s getting a lot of pleasure out of ‘bachelor’s handbag’. Are you aware of the term “bachelor’s handbag?”
Mark: I’m not. I’m so out of the scene, clearly.
Daniel: Go to the grocery store and there’s cooked chickens in a plastic bag.
Mark: Cooked chickens in a plastic bag?
Daniel: You know, the kind of chickens…[crosstalk]
Mark: Yeah, I know the ones.
Daniel: You can just eat them. You don’t have to cook them, you could just take them away. That’s a bachelor’s handbag.
Mark: Really? [laughs]
Daniel: So, there’s your metaphor.
Mark: Does it have to have roast chicken in it, or is it any plastic bag?
Daniel: It has to have roast chicken in it.
Mark: That’s hilarious.
Daniel: Because it looks like a handbag and you’re carrying it around and you’re going to take it home for dinner. So, that’s your metaphor. It pulls in a lot of fun things about like handbags and the inappositeness of putting the term ‘handbag’ onto a ‘bachelor’, because then usually so what could it be, so that brings in a lot of domain knowledge about what bachelors do and how they don’t like to cook food and they’ll just take it and eat it.
Mark: That’s very funny.
Daniel: Yeah. People do get a little rush. “Oh, that’s good. I’m going to start using that.” Very often, we’ll have people say, “I just learned this term and I’ve got to work it into a conversation.” So, you can really see how new things come in and get taken up. And people like that.
Mark: People do. They love it. Think about scribes, the first time we thought about using pictures to represent sounds, by the first sound in the word, we draw something that has that sound as it… sorry, I’ve said this the wrong way around. We draw a thing for which that sound is the first sound in the word.
Daniel: You have aleph, and pretty soon it doesn’t mean aleph anymore, it means ‘ul’.
Mark: Exactly. That’s a wonderful example. We can see this. If you look back in the history of writing of our alphabet, the Roman alphabet, which derived from a version of Greek alphabet, derived from a Semitic alphabet, and then related to Hebrew alphabet, then if you look at the names for the letters and what those letters represent, the names are things in the world whose first sound is the thing that’s represented by the letter. So, aleph. You know what alif was…[crosstalk]
Daniel: It’s an ox.
Mark: It’s an ox, exactly. And beth.
Daniel: Which is a house.
Mark: B, so beth is house because the Hebrew word or the Semitic word at that stage, and bayt in Arabic and beth in Hebrew.
Daniel: Bayt in Egyptian.
Mark: Exactly, because that’s Arabic.
[laughter]
Mark: So, this is where our alphabet startup, was using basically part of the sound of a word, then representing it with the picture for that word, corresponding to that word, something really physical that you could draw easily and then using that to represent that sound. It’s not just a one-off thing. This happened for Semitic, but we also see the thing in Egyptian hieroglyphics that come to mean single sounds have complex, really pictorial pictures, not very stylised, but not abstracted, but you can still recognise the eye’s an eye.
Daniel: It’s like the sandal strap, which was something like unk. They started using that for the word ankh which mean life. It’s like, “Oh, I can pick that up because it’s easy to drop. And I can use it for other stuff that sounds the same. That’s really cool. I’m totally going to start doing that.”
Mark: This happens in Chinese as well, completely separate history but they start off making these very pictorial characters. They get very stylised quite quickly. They say, “Well, we’ve got all these words that sound alike. Why don’t we use a bit of the symbol for the meaning, the general class of meaning, combine it with something that is a word that sounds like it, represents a word that sounds like the thing, even though it means something completely different, sounds like the thing?” This combination of like, “Okay, now I’m talking about a watery thing, and it sounds like joel.” You put those two together and it tells you whatever a watery thing that sounds like joel is. I don’t know Mandarin enough, I should have got a natural example but let’s say it could be river, for example, or something like that. But the idea is you get this combination of something that tells you roughly the meaning area that you’re in, plus something that tells you about the sound from words that just coincidentally sound the same. Those two things combine to give us the modern general class of Chinese characters.
Daniel: People take shortcuts, just like the Pictionary game.
Mark: It is really clever, like the Pictionary game.
Daniel: Now, we’re ready to give a more full picture of what we’re doing right now. We have a pile of stuff. We’re trying to add to it, but while we’re doing that, we are making predictions. As I said before, we’re making predictions about what the other person is going to say. We’re making guesses about each other’s knowledge, beliefs, desires, intentions, plans, and goals. But we’re also, word by word, making predictions about what word is going to come next.
Mark: We are.
Daniel: This is, by the way, independent of the idea that we launch into sentences without planning them fully. We don’t know what we’re going to say. I don’t know how the sentence is going to end it, but I just have a general idea and I’m going like three or five words in advance, but you’re listening to me like three or five words in advance, making predictions about what I’m going to say. If that doesn’t seem to match, there’s got to be a reason why either I’ve misunderstood or I’m trying to highlight your attention… I’m trying to control your attentional focus onto something because I think it’s important or it’s going to defeat a wrong reading that you’re currently having or something.
Mark: Exactly, or something but those two are the main ones. It could be defeating like a wrong belief you actually hold, or else helping you avoid a misdirection, like helping you avoid going down the wrong path in interpretation.
Daniel: And all of this helps make conversation more efficient because frankly, there is a lot going on.
Mark: There is so much going on. We also need to have our attention focused. Our attentional brains, our conscious minds have a very low throughput. They can’t cope with so much.
Daniel: No, not for too long, anyway.
Mark: Not for too long. Actually, I just discovered, I think… was it yesterday I mentioned this to you, about seven plus or minus two?
Daniel: Yes. George Miller’s observation that we can only hold like five or seven or nine things in our conscious attention.
Mark: Exactly. By which he meant kind of a short-term operational memory, but I think it’s also a good way to say is that conscious in the moment attention, he said that you break down nine things, but you might also break down at five, just depends on the things, the circumstances. This original observation has been modified in many ways by psychologists looking at different aspects of cognition, but it still roughly holds true. We can pay attention to not many things at once.
Daniel: I’m constantly becoming aware of the ways in which these attentional shortcomings that we have or these deficits in working memory really shape how language is. You’re going to keep dependencies close together, like you’ll have red house or house red, but you’re not going to have house entire phrase red at the end of it.
Mark: This is something that one of my favorite observations in about English, which I would argue if I ever to dip my toe into syntax in English, I would have a very functional theory of syntax that would say a lot of English word order is motivated by one principle, which is keeping related words, words that have related meanings, like selectional restrictions, keeping those close to each other. We see this in something called dative NP shifts.
Daniel: Okay, hang on. “John gave the pen to Mary.” That’s dative. “John gave Mary the pen.”
Mark: Exactly. What we…
Daniel: [crosstalk] -shifted.
Mark: …the pattern in English quite strongly is the one you put closest to the verb should be a short one.
Daniel: Oh.
Mark: Because think about this. If you want to have words or phrases as close to each other as possible, if they are related, the problem is your verb is related both to the direct object, the thing that’s being given…
Daniel: The pen.
Mark: …the pen. And to the indirect object, the recipient, the one who’s getting the pen. They’re both related to the verb.
Daniel: Just in different ways.
Mark: Just in different ways. So, they both want to be next to the verb, but they can’t be.
Daniel: [chuckles] Yeah.
Mark: But if you put the shorter one closer to the verb, that means that the distance before you get to the other one is less. So actually, you’re solving this problem, you’re making this difficulty better alleviating it if you put the smaller one near the verb.
Daniel: I have an example of this.
Mark: Tell me.
Daniel: I could say, “That gives me the creeps.” But I can’t say, “That gives the creeps to me.”
Mark: It would be silly.
Daniel: But if it’s a very long dative, then I can do it. I can say, “This movie would give anyone who’s even slightly nervous the creeps.” Oh, I’ve forgotten what the thing was. “This movie would give the creeps to anybody who…”
Mark: Perfect example. That’s just trying to keep the related words close to each other, the related phrases close to each other.
Daniel: Because otherwise you forget.
Mark: You forget. Yeah. We think of this as kind of dragged out as like an oddity in English, but actually I think it’s a fundamental principle that organises so much. For example, English is a subject-verb-object language. If you have a subject and an object, not two objects, but just one object and a subject, and the verb is related to both of those, it’s like the thing that relates those two together, then the way you get things closest to their related words is by putting them in subject-verb-object order. You could have object-verb-subject as well. But that’s the only two options.
Daniel: Yeah.
Mark: There’s other reasons why the subject goes first but…
Daniel: We’ve done a whole show on that.
Mark: I would be surprised if you hadn’t.
Daniel: Is that because subjects are usually here are old?
Mark: Exactly. They are a way of easing yourself into your information.
Daniel: We say, “The sun arose on the city,” so now the city is a known entity. “The city was beginning to stir,” I put that first because now it’s… [crosstalk] I go from old to new.
Mark: Exactly. That’s many ways in which that is a really good cognitive approach to… because one of the things your listener is dealing with is, are you going to start a new sentence? What thing could you talk about? At the start of a new sentence, there’s a lot of unknowns. If you start with old information, you reduce the strain on the listeners processing and then you can gradually introduce the newer stuff through steppingstones of relationships.
Daniel: There was once a man, and his name was John. Now, he’s known. John got up this morning and petted his dog. The dog had been asleep for hours but now…
Mark: I think there’s many situations where we actually have ritualised starts. For example, to discourses of particular kinds or fairy tales have a standard opening in English.
Daniel: Yeah. Once upon a time.
Mark: And this reduces its…
Daniel: …cognitive overload.
Mark: Exactly. Anything could happen at the start of a story. Think about the first lines of novels and there are so many first lines you can have in a novel or first words.
Daniel: You just dropped in.
Mark: When people do a great job, we get really impressed and we remember those like it was the best of times, it was the worst of times.
Daniel: Mother died today. [crosstalk]
Mark: Exactly. [chuckles] There’s so many times people have done it well, but there’s also a lot of boring times where it was just confusing and were just not memorable. But if you have a set phrase, you get past that trouble.
Daniel: If it’s a really known set phrase like, “It was a dark and stormy night,” now you’re not just doing something that’s known, you’re actually tying it into a body that helps people know what to expect.
Mark: That’s so true.
Daniel: I’m getting the picture that a lot of conversation is the managing of expectation.
Mark: A lot of it is, I think a lot of the work we do and there’s really good reasons for this. I’m not sure I know all the reasons but let me tell you about one that I’ve been thinking about a lot recently.
Daniel: Okay.
Mark: If you are trying to communicate efficiently with someone, then we think, “Okay, we don’t need redundancy.” Our intuitive idea is that we should eliminate redundancy to speak efficiently…[crosstalk]
Daniel: Many people feel this way.
Mark: …with someone and that is really perfectly fine so long as your communications channel is absolutely perfect. The other requirement is that the space of things that could be talked about is well known.
Daniel: I already have to know everything you know, and we have no problems with our communication channel.
Mark: [laughs] Exactly. Those are the two requirements.
Daniel: Efficiency is fine as long as you already know what they’re going to say.
Mark: Yeah. And then, you can be perfectly efficient. We have ways of organising for exactly that circumstance. For example, when thinking about coordinates for maps where we minimise redundancy, so it’s like you’re on page 34 of your street directory and A7, if you remember street directories or these GPS coordinates, if you’re just someone that is cruel enough to give you their home address by GPS coordinates, do they actually want to see you? I’m not sure.
[chuckles]
Mark: But they are perfectly… what’s the word? Nonredundant representations of that information.
Daniel: But we don’t think that way.
Mark: We generally don’t think that way.
Daniel: We don’t think that way. We don’t communicate that way.
Mark: Well, we don’t communicate that way for really good reason and that is partly because we don’t think that way. We have someone who’s interpreting those things is very likely to make a mistake because we don’t process very well that sort of information. It’s where we can process that in a kind of very formalised way, then we can use these efficient codes. We have lots of experiments in psych, where we look at how people shift towards using efficient codes away from using more redundant codes as they become more… what’s the word? Skilled at the interpretation and they can do things more efficiently, like we talked about earlier with the Pictionary task.
Daniel: Yeah.
Mark: They get more and more efficient in their representations.
Daniel: You’re working together on the same project, you’re approaching the same pile of knowledge in the same way.
Mark: Exactly. So, this is the situation. And usually for those situations, we try and construct perfect communication channels for them. But as I said, communication is very noisy, particularly if the person you’re talking to is not paying very close attention, which is the normal state of affairs in communication.
Daniel: The same thing could happen if our brains were somehow better, and we had more attentional focus.
Mark: Exactly. So, if we have…
Daniel: A better working memory.
Mark: Think about it this way. If I emphasise my words and speak very clearly, then I do not need to say so many words. I can say one word to name a thing and be more confident that you have heard me. But it’s annoying to talk about that. It’s annoying to hear that. So, it’s probably slightly better if I just, say, add a few more words and say them in my normal slovenly manner.
Daniel: I had an actor friend whose name was also Mark. One day, I met up with him. I said, “Hey, Mark, how’s it going?” He said, “I’m fine. How are you, Daniel?” He was doing this. I realised because he was studying drama, he was probably doing an exercise and that was fine. I went with it for a while. But then after a while, I wanted to strangle him because it was highly annoying. It simply was not necessary for him to be speaking at that level of clarity because we can strike a balance. We’re always striking that balance between how much effort we’re prepared to put in and how much slop we’re allowed tolerate.
Mark: This is what redundancy helps us do. If we didn’t have redundancy in our communication, we wouldn’t have the option to be lazy because we’d have to be constantly paying perfect attention. A, we can’t do that, and B, we don’t want to.
[laughter]
Mark: We so don’t want to.
Daniel: So, it is a little bit lazy.
Mark: It’s lazy. It’s very lazy in a good way in its expectations of laziness. For example, we were talking earlier about how do we design our conversation for an audience where there’s lots of people we don’t know. We don’t know exactly what they know and what they don’t know and what their attentional situation is in the moment that they’re listening to this. But what we can do is we can have a fair amount of redundancy in the way. We speak and if they lose focus for a moment, they won’t have to rewind to hear our dulcet tones and say the magic words again because there’s enough redundancy, they can pick it up. We all have this experience of like we zone out for a moment in the conversation and come back.
Daniel: [laughs] I’m thinking, is the camera still working?
Mark: [laughs] Exactly, that sort of thing. But when we come back, we can piece together enough that we’re happy to continue without actually rewinding life to…
Daniel: Without having to go back and do repairs. “Oh, my gosh. I didn’t catch what you said that one time.” “So, boring. Oh, my gosh.” That’s inefficient.
Mark: Exactly. I think that this redundancy helps us with the fact that the channel is imperfect, and we vary the amount of redundancy in order to help our listeners. And we know this. There’s something called the… something like constant information rate. It’s not the exact wording, but it’s something like that in phonetics where people have studied how much information is actually being conveyed by the words.
Daniel: Oh, yeah. And it’s the same across languages.
Mark: It’s not just the same across languages, but even as they’re saying more informative things, they slow down.
Daniel: Yeah.
Mark: When they’re saying less informative things, they speed up. Within one language, within one speaker, within one conversation, they’re adapting to how informative what they’re saying is so that the listener has a fairly even ride but isn’t going on a smooth road, then hitting a speed bump of vital new information that was said in half a second. But instead, it’s stretched out with redundancy whereas you can reduce the redundancy on the predictable things because you don’t need so much.
Daniel: We just said that we’re a little bit lazy and a little bit unfocused sometimes. When we talk about it like this, we realise no, our brains are just churning through data and arranging things optimally under the surface when we’re not even aware.
Mark: Yeah, doing a lot of work.
Daniel: How is it that I’m able to parcel out the information that I’m giving in a mostly constant rate, even when the topic gets more difficult or the purpose of the conversation is different? That’s weird.
Mark: That’s weird. And it’s amazing.
Daniel: [laughs]
Mark: I think going back to that previous metaphor I gave, it’s like a cricketer who’s constantly having cricket balls flying out of the deep sky at them, and he’s catching them one after the other because he’s continually making these predictions and then catching the ball. New one coming, catch the ball after prediction.
Daniel: We’re good at this.
Mark: We were so good at this, but it involves attentional management so we don’t get exhausted. I can give you first-hand accounts of this, and I’m sure you’ve probably had this experience too, Daniel. Many of our audience will as well. You’re learning a second language. You spend some time talking to people in their second language, you get exhausted. You absolutely do, because you can’t afford to be as lazy, because you can’t make predictions.
Daniel: You can’t make predictions. You don’t know the transition probabilities from word to word.
Mark: Exactly, and from meaning to meaning and sentence to sentence and phrase to phrase.
Daniel: And you don’t know the selectional restrictions. Like, you weren’t able to take advantage of all the redundancy that was in there because you were unaware of what it even was.
Mark: Exactly. When we learn words, we learn just the core meaning of the word. They, as you said, don’t tell you about the selectional restrictions. You don’t learn that as part of the package while the native speakers do. They have enough experience to have learned that and then they can use that to help them make predictions.
Daniel: And you’re only going to acquire it unconsciously through experience.
Mark: It’s so annoying. [Daniel laughs] But this relates to actually one thing that a friend of mine who’s had a stroke this year and is trying clawing back his language facility again. Fortunately, he’s able to communicate well, but it’s slow. One of the things I was suggesting to him, where I think that maybe the damage has hit some portion of his production, is in terms of this sequence probabilities. His words are chopped up into much more deliberate segmental expressions or syllabic, saying one syllable and then the next syllable as if they’re quite separate. I suggested to him, in as far as you’re able to read, he’s able to read actually enough to pronounce. Making sense of what he’s reading is harder because of the stroke, but I don’t think he needs to do that. I think if he can just read aloud text and practice that, then that will help him retune other neurons with the brain plasticity we know lasts all our lives to help build back this facility to predict what’s coming next just in terms of the syllable goes after the syllable, which is actually quite a big help in terms of the sequencing of words in English.
Daniel: Co-ffee ta-ble.
Mark: Exactly, that sort of thing. By just exposing yourself to lots of sequencing of sounds, you will over time internalise that and that will, I think, smooth production for him. Make it easier so that his unconscious… It may not make sense off on what syllable he starts saying next, but there’s a good chance. If he’s exposed to a lot of English and he has this predictive capability back, even if it’s a dumb prediction, it’s reducing the load on the conscious mind that’s having to construct it. So, the conscious mind is now choosing between a smaller set of options for what comes next.
Daniel: It’s time to summarise. What have we got here? What I’m getting is that predictability matters, and redundancy helps.
Mark: Absolutely.
Daniel: We are making predictions and we’re smoothing out the way for our listener whenever we can.
Mark: Exactly.
Daniel: Language also looks the way it does because of attentional lapses that we make… the amount of attentional effort we’re willing to expend, and also just working memory limitations, those influence the way language looks.
Mark: They really do.
Daniel: When we’re trying to help interpretation along, we have a bag of tricks that we can use to make things more prominent, to adjust the rate and flow of information to overcome deficits of a noisy channel, and we do it all automatically.
Mark: We do it without thinking automatically and incredibly fast. We make all these predictions, and we negotiate our predictions about what’s going on in this combined mental state picture we have of our conversation. We process that and slide seamlessly into the right choice for what to do next.
Daniel: You are figuring out how it all works.
Mark: Yeah, that’s my plan.
Daniel: [chuckles] Very cool. This is a conversation that goes on… This is just a short part of a conversation that’s been going on for about a week between me and Mark, pretty much all day long, [both laughs] which will continue until Mark goes home. [Mark laughs] But I’m glad we were able to get of this stream so that we could talk about your work. Thanks for hanging out and thanks for talking about it. It’s really fun.
Mark: It’s been lovely. It always is, Daniel. It’s so good that we could have this longer conversation.
Daniel: There you go, folks, a deep dive into language, all of it. I enjoy these deep dive episodes. I hope you do too. Thanks to Dr Mark Ellison for coming and having a chat.
And thank you for listening. You can get in touch with us if you want to. We’re @becauselangpod on all the socials. You can send us an email, hello@becauselanguage.com. You can also become a patron and support the show as little as a buck a month. Gets you onto our Discord with all the other Discordians at Because Language. Bonus episodes of the listener level.
I’d like to give a quick shout out to our top patrons: Iztin, Termy, Elías, Matt, Whitney, Helen, Jack, PharaohKatt, Lord Mortis, gramaryen, Larry, Kristofer, Andy B, James, Nigel, Meredith, Kate, Nasrin, Joanna, who for this episode is bumping it up to the Supporter level after years of being a patron. Thanks, Joanna. Ayesha, Moe, Steele, Margareth, Manú, Rodger, Rhian, Colleen, Ignacio, Sonic Snejhog, Kevin, Jeff, Andy from Logophilius, Stan, Kathy, Rach, Cheyenne, Felicity S, Amir, Canny Archer, O Tim, Alyssa, Chris W, Felicity G. Our newest patron at the Listener Level, Vee. Thanks to all of our amazing patrons. You can also donate to us on Ko-Fi… Ko-Fee, Ko-Fee in case you don’t like Patreon for some reason.
Our theme music has been written and performed by Drew Krapljanov, who’s a member of Ryan Beno and of Dideon’s Bible. Thanks for listening. We’ll catch you next time. Because Language.
[Transcript provided by SpeechDocs Podcast Transcription]