Conversation Analysis is a powerful tool that looks at large numbers of conversations to help build insights about what works and what doesn’t.
Elizabeth Stokoe is a Professor of Social Interaction at Loughborough University, and shares some key insights from her excellent book, Talk, the science of conversation and her well-received TedX talk.
As she suggests in the opening quote, any conversation that you participate in has a landscape to it.
What Conversation Analysis can do - and we are all conversation analysts, just not professional ones - is show us the texture of that landscape, and how to navigate the bumps in the road effectively.
One surprising idea I absorbed from Professor Stokoe is in this quote, when she says that:
“In a way, the best conversations might have some clumsy, awkward moments and through that way, you might move past it and into something more mutual”
We know what is natural and easy because we know what feels clumsy. Seeing, accepting and moving past the clumsy can help us find a smoother path.
We are the Turns We Take
Elizabeth’s idea that we are the turns we take, that speech acts are real acts, is a powerful one. And so is her idea that non-responsiveness or silence in reply to an awkward turn can get things “back on track”. If someone comes in “hot” to a conversation an easy way to cool things down is to wait and let the person fix it themselves, as she says:
“People will figure out that they just did something that was a bit off and fix it.”
What I really loved about talking with Professor Stokoe is that she busts conversation myths with ease - and Science!
There are many popular ideas about conversations, from how they differ across cultures to how much communication consists of body language to how men and women speak differently - both in amounts and type.
Professor Stokoe suggests that there are many more similarities than differences across cultures and genders. She is in fact, more interested in how we construct gender through speech, than how our biological gender influences speech.
And she also reasonably suggests that if body language is 90% of communication, why can we communicate just fine over the phone? There is, as it turns out, very little science to support many such figures.
Working with real conversations instead of simulations
Elizabeth also casts very reasonable doubts on some of industry’s favorite models to explore interactions, like secret shoppers - it turns out that people who are acting like customers don’t act like customers.
She also suggests that using role-play in training is not as effective as it could be.
Conversational Analysis can offer better insights by studying real conversations en masse, in fine-grained detail.
Be sure to listen all the way to minute 45 when we dive into group conversation dynamics and how people learn what behaviors are acceptable in a session in the opening seconds of an interaction. It is shocking how quickly the landscape of a conversation is built and surveyed by the participants.
Links, Notes and Resources
A deep dive on her work on the TED blog
Elizabeth’s excellent book, Talk
On Body language:
https://en.wikipedia.org/wiki/Albert_Mehrabian “Mehrabian's findings on inconsistent messages of feelings and attitudes (the "7%-38%-55% Rule") are well-known, the percentages relating to relative impact of words, tone of voice, and body language when speaking. Arguably these findings have been misquoted and misinterpreted throughout human communication seminars worldwide”
Lenny the anti-cold-calling chatbot
More about conversation and gender from Professor Stokoe here.
Full Transcription
Daniel Stillman:
I'll officially welcome you to The Conversation Factory, Liz. Thank you for making the time.
Elizabeth Stokoe:
Nice to be here.
Daniel Stillman:
Thank you for saying so. Let's start at the beginning. What features of talk are universal?
Elizabeth Stokoe:
It's a great question and it's a question that people ask regularly of someone like me, a conversation analyst. A couple of ways of answering this question. I think the first way is to think about what we're doing when we're interacting, when we're talking. What we're doing is we are building and responding to actions, so as soon as you start to think about social interaction, talk, conversation as actions being built, progressed, halted, suspended, moving forwards and so on, then you start to realize that in fact, there are lots of universals across cultures and across languages. For example, questions and answers, they are massively constitutive of a lot of our daily life and all languages provide for us to ask questions and answers. I think when you start to think about what people are doing when they're talking, they are asking questions, requesting, assessing, evaluating, greeting when languages provided for those things.
Elizabeth Stokoe:
That's not to say that languages provide for different ways of building actions, but nevertheless, the core underpinning universal of human sociality is that we are doing actions when we interrupt each other. Then there are some empirical studies over the last 10 years or so, where people are explicitly comparing different languages and how they are building different courses of action. For example, turn-taking, which is one of the most fundamental things, again, a universal thing of a conversation. There's some research by Tanya Stivers and colleagues, where they compared lots of different languages from around the world. They looked at the timing of turn-taking and surprisingly, universally, at least in terms of the large study that they did, they found that each turn lasts for about two seconds and the typical gap between them is about 200 milliseconds, regardless of grammar, language, and so on.
Elizabeth Stokoe:
We tend to assume that cultures are all different and they very well might be. That's the question for other types of analysts, but in terms of social interaction and conversation, we can start to see that there are more similarities than differences when we think about action. I think one final way of thinking about this question is, this isn't technical at all, but it's something that I think about quite regularly, which is something like the interactional imperative. I can remember my grandmother when she became really old and hard of hearing, she nevertheless knew that turn-taking must still happen, so she got really good at doing mm-hmms and nods, and little things that sustained the interaction at the right place, mostly. Occasionally, she might get it wrong, but that sense of an interactional imperative, that one must keep taking turns and keep on producing social life through the machinery of interaction, I think is universal.
Daniel Stillman:
This idea of conversation or talking as a project and not replying to the offer … There's this fundamental judgment of, the person is uncollaborative. It's just interesting that there's this imperative to not look like an unhelpful person, to not be that sort of person.
Elizabeth Stokoe:
It's through the fact that we know that that turn-taking just keeps on tricking, but we can see we have a basis for describing somebody as uncooperative or not collaborative.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
When we make assessments of people, we're not generally making psychological assessments as a psychologist. We're basing or our ordinary diagnoses of what people are like, how rude or lovely-
Daniel Stillman:
Yes.
Elizabeth Stokoe:
What people are doing. A lot of what people are doing is what they're saying.
Daniel Stillman:
I mean obviously, we're going to spend the rest of our conversation unpacking all of that because in your book, which is really, really lovely, you have this phrase, “You are the turns you take.” I think this is a really wonderful idea and you've talked a little bit about it just now. Can you say more about what this means to you, that we are the turns that we take?
Elizabeth Stokoe:
I just think it's really interesting. I'm a psychologist by background. Psychology owns the domain of understanding human beings, but most of us aren't psychologists, and most of us, as we proceed along and encounter, suddenly halt it, step outside it, do a psychological assessment of it.
Daniel Stillman:
There's no survey that we're checking off. There's a diagnostic. Here's the four-part diagnostic we're completing.
Elizabeth Stokoe:
We're not doing that. We are experiencing interaction in real-time and then, later on, maybe talking about that interaction to somebody else, and basing a lot of our assessments of people on the basis of what they have said. I don't want to make a distinction between what they've said and what they've done because I want to say that saying things is doing things.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
On that basis, we are the turns we take. This becomes really important in places of application of my work. It might feel a bit of an extreme instance of this, but we know it to be true, which is that when you look at crisis negotiators talking to people in the most extreme, maybe awful extreme moments of their lives, then this crisis negotiator, no matter what we might think, is structuring that in that encounter and what might be motivating the person who maybe stood on a roof or threatening themselves or somebody else. We tend to reach to understanding that situation in terms of the individual psychology of the person who is in crisis, but we don't have all of that information when we encounter them.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
When a negotiator encounters that situation, they can't get a psychological assessment of that person. All their evidence of how to develop a strategy, if you like, is based on what the person says. That is what we're doing all the time.
Daniel Stillman:
What they're not saying potentially.
Elizabeth Stokoe:
What they're not saying, yeah.
Daniel Stillman:
This really comes back to this. I've got my notes here. I mean we need to unpack what a conversation analyst is, but there's also this idea in your book, that we are all conversation analysts. We're just not doing it with the scientific rigor that you do. I would love for people to understand. I mean your book is like zooming in minority report style, really just exploding moments of conversations and doing a very, very fine reading of them to say what is happening at each one of those moments. We're also all doing that. You give this example in your book, I think it's hilarious how much pop culture is in your book, of Olaf in Frozen noticing somebody's response being a little slow, saying, “You hesitated.” What does that mean? We're always reading people's intentions.
Elizabeth Stokoe:
That's right. Basically, we are all analyzing all the time. Otherwise, we would never be able to take a turn because mostly, when we are in an encounter, we are listening for the actions being done by the person we're talking to or people we're talking to and finding a moment to respond, which can be difficult. We all know what it's like to crash and talk at the same time.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
We are analyzing for the action, so if you're assessing something, then I'm going to either agree with you or disagree with you, and I'm going to have to manage that or move with it. If you say something and I don't understand. Then I'm going to have to display that and we can go really maximum on, “I don't understand you,” by articulating that. Or we can just say, “Huh” or just a little delay, and sometimes that little delay is enough for you to realize she didn't get that and do it again. We're monitoring ourselves all the time as we build courses of action in conversation.
Daniel Stillman:
It's funny. It seems exhausting when we explain it that way and yet, it's the thing that, I wouldn't say we're built to do, but we've built ourselves to do it in a way.
Elizabeth Stokoe:
Absolutely. I mean I use that idea actually to think about this book, which was designed for a non-academic audience-
Daniel Stillman:
Yes.
Elizabeth Stokoe:
At least to-
Daniel Stillman:
Thank you so much, by the way.
Elizabeth Stokoe:
To show conversation analysis to lots of different industries and so on because standing on a stage and doing a popular science of conversation can be a bit of a challenge. Everyone talks, so everyone thinks they know about talking already. There is so much pop-psych out there, about communication. That everyone has a view, a very strong view sometimes, about what they think the world of communication is like.
Daniel Stillman:
We must because we get through our days and we get through our lives, hopefully with a reasonable amount of success, yet-
Elizabeth Stokoe:
As soon we go meta on communication, then people's myths and stereotypes start to pop out and they may or may not have any basis in what a conversation analyst at least would take to be evidence of a robust kind.
Daniel Stillman:
What are some of those myths that you think are important for people to step back from?
Elizabeth Stokoe:
My favorite one is the body language myth.
Daniel Stillman:
Oh yes. The 93% that people wrote about.
Elizabeth Stokoe:
Some very high percentage of communication is body language versus some other much lower percentage, which is the words uttered. What I like about that is it's so, so massively perpetuated. I've sat in loads of events where again, it's not necessarily an academic event. It's a training event. It's some professional event and a speaker might get on and do something before me, and that slide will pop up. It's actually an interactional challenge. Do I challenge it or just let it go? I mostly just let it go.
Daniel Stillman:
Why? Because you're just so tired of fighting that fight?
Elizabeth Stokoe:
Yeah, but the interesting thing about that one is that it's very easy to knock it over. There's a conversation analyst called Max Atkinson who's one of the few people who has actually taken the conversation analysis outside of the academy and into other domains. He basically interviewed the author, Albert Mehrabian, I think he's called, of the original article that found that back in the 60s and found something that got used and turned into that stat. I think not only did Max Atkinson argue that Mehrabian himself had joined the campaign to stop people perpetuating that particular myth and get communication trainers to stop using it. He said something like, “If this was true, then none of us would need any other language than body language, 90% of the time.”
Elizabeth Stokoe:
We wouldn't need any other languages. We'd all be able to operate in the world. Of course, that doesn't make sense. Then we wouldn't be able to talk in the dark or on the phone. As soon as you start to think about it, you think that can't really be right, but yet, it persists a bit like almost everything we also do know about language. The simpler the message, the more soundbite it is, the more compelling it is, but then the more difficult it is to challenge it when it's just passed out of the realm of any article back in the 60s. It's just what people think they know about talk.
Daniel Stillman:
That's very true. I feel like narrative is an element of talking, of conversing with each other. What we're sending is messages and stories, back and forth with each other and we're building a story together. I think this goes towards the racetrack concept in your work. That we know the general shape of the project that we're working on. There's summons. There's an exchange of greetings. There's question and response. We know the vacuum that's formed when somebody asks a question and there's no answer. In a way, how does that …? What's my question here. The racetrack is imprinted on us. How can that concept of the racetrack help us in shaping our conversations for the better? I guess that's maybe the best I can do with that thought.
Elizabeth Stokoe:
I think the reason that I started to use the analogy of a racetrack, or I might even say a golf course or anything with an architecture landscape to it, with hurdles and possibilities for falling over, mostly what I wanted to communicate is this idea that any encounter is built of projects, be they small or large things. Any conversation that you have on the phone with a salesperson or the council or the doctor, it has a landscape to it. You move through a series of things that need to get done before you can move on to the next one successfully, at least. If you phone up the doctors to make an appointment, then there'll be a bit of identity. There'll be the opening bit where you realize you're both talking to the right person that can make the request. Then there might be some complication around progressing the request.
Elizabeth Stokoe:
Hopefully, the request will be met and then you confirm what's going to happen, and then you close the call. The fact that these things, when you look at lots and lots of them, they do tend to have a macro organization like this kind of landscape. It allows us to see things out of place really quickly. When you are on a call where you feel like you're being pushed to make a decision, for example, you're probably feeling that because it's been moved up the racetrack. That they've got some point which should really fit further down the encounter. We might have a sense of that and a conversation analyst would probably be able to show that when you look a batch of conversations. The racetrack, first of all, gets us to think about, everything that you do creates the possibility for the next thing, every single time.
Elizabeth Stokoe:
It can be probably quite worrying. You really start to think about it. Everything I say is making it possible or easier or less easy for you to do something next, every single time. I'm constraining or restricting what can happen. That's something. The other thing is to get people to understand that social interaction is much more organized and not messy. People tend to feel talk is messy. It's why we do all sorts of other things to understand communication than bother to study actual conversation because there's, again, this other myth around, talks are really messy. We have to produce idealized versions of interaction to understand that. I guess 50, 60 years of conversation analysis shows that that's just not true, that the interaction is highly organized and yet, people can still be idiosyncratic within that organization.
Daniel Stillman:
I think at least one thing I want to pull out is this idea of when things are off track, we have the sense that things are off track and many people will have to deal with uncollaborative situations or uncollaborative conversation partners. What can you tell us about how, if things are off the track, to bring them back on the track? I know there's examples like Gordon restarting the conversation. He just tries to bring it back on to track, but those ways, to me, feel clumsy and a little awkward.
Elizabeth Stokoe:
I mean clumsy and awkward is part of everyday life. We need clumsy and awkward just so that we understand what isn't clumsy and awkward as well. In a way, the best conversations might have some clumsy, awkward moments and through that way, you might move past it and into something more mutual. I think it's interesting that you raised that. The Gordon example you referred is the start of a telephone conversation between Gordon and his girlfriend on the phone. She breaks the start of the racetrack straight away by asking … Rather than saying, “How are you?” “Fine.” “How are you?” She says, “Where have you been all morning?” It's noticeably out of position, if you like, in the normal, unfolding racetrack and we all see it because we all tacitly without ever thinking about it before right now, know that what belongs there is, “Oh, hi.”
Elizabeth Stokoe:
“How are you doing?” “Fine. How are you?” Then you move on to the next thing and she replaces that with something else. What Gordon does in response, rather than answering her question about where he's been all morning, the first thing he says is, “Hello,” which belongs there. It might sound a bit disingenuous, especially the way I reproduce it now, but I think there are some things to be learned from that, which is that if you just understand that you can always resist the thing that is being set up for you in the previous term, it might make it clumsy and awkward. Sometimes what you're also doing is a little bit of socializing of people. If you can, then … It can be hard to do this, but I guess I would try to do that socializing sometimes if you're in a position to do it.
Elizabeth Stokoe:
If you walk into a meeting room and someone doesn't say hello or they don't know who you are and they just say something that is a bit off, then you cannot respond to that in the terms set up by that person. You can just say, “Oh, hello” and just do the thing that belongs at that point in the encounter. A lot of the time people will get the message. What you can do is just stay silent, actually, give a slightly blank expectant look. Very often, people will repair things themselves. They'll get a second, a tiny fraction of a pause really. People will figure out that they just did something that was a bit off and fix it.
Daniel Stillman:
It's a reset and literally putting things back on track, and saying, “Oh, good morning.”
Elizabeth Stokoe:
I think this is one of the things that, again, you might think that if you are going to call someone out in any kind of way, that you need to explicitly call someone out, whereas what we see interaction is a bit of silence. If you are face-to-face with a slightly blank, expectant look like, “Oh, I'm tricking you. We're not done yet. Let's see if you fixed it yourself,” that will often work.
Daniel Stillman:
This is escalation versus de-escalation, I think. Responding directly like, “Well, that's rude,” is taking the turn and then one-upping it versus taking the turn and one-undering it.
Elizabeth Stokoe:
I think saying, “That was rude,” at least again, it's this talking theory, which is I had to talk in theory. Then when you look at real interaction, you see how absolutely difficult it is to say, “That is rude” to someone, directly.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
I mean your heart will be beating if you did that. Yet, we're full of advice to people. Just tell them this, but it's actually really difficult when you're in an interaction. I think a really nice example of this is looking at … I've done some research with colleagues on cold calling, not people calling people's domestic spaces, but just business to business. Can we make an appointment to show you whatever it is that we're selling? In that collection, there's a very, very small number of hang-ups, despite the fact that most people will say things like, “Oh, I just hang up on cold callers.”
Elizabeth Stokoe:
Of course, whether they do or not, right now when they're saying that to whoever it is they're talking to, they're doing something with that now anyway. Whether they actually hang up on cold calls isn't really the point. They're just showing like “This is the kind of person I am. I am telling you this to assert what I would do in this situation,” but we actually see very few hang-ups. People are desperately trying to get out of the call.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
It's pretty unusual. People have to be pushed quite far by a salesperson to hang up in empirical reality, at least.
Daniel Stillman:
Well, I mean, I think this is worth unpacking as well. I'm really curious about the application of conversation analysis to training and behavior change because as you report, we don't often know what it is that we're doing. We don't. Actually, we're not noticing the placement and order of words that you are noticing when you look at the actual transcripts and when you're looking at hundreds of transcripts across multiple conversations. You're looking at a histogram. You're looking at patterns. I may not even know what it is that I'm doing, that is effective and so it's very hard sometimes for me, if I'm a great salesperson, to teach someone else how to be great like me at sales.
Elizabeth Stokoe:
Absolutely. At least the first hurdle there is that people do have a view about themselves or others. Also, I think they know, just by inspecting their memory, what may or may not work in an encounter. I think I'm going to assert this strongly and say people really don't know what they're doing.
Daniel Stillman:
No. Because we're doing it, it's very hard to notice what you're doing and do it at the same time.
Elizabeth Stokoe:
Yes, even though they are tacitly, maybe doing a fantastic job as well. We know that people have great skills at interacting with other humans. At least some of them do. Some of them don't but again, because communication is just a thing that everyone thinks they know about. The first hurdle is getting people to pause and think, okay, if we are going to develop training or guidance maybe or especially assessment of other people's communications skills, then how best to do that. I think I love the example of that, some research conducted with my colleague, Rein Sikveland. We're looking at … Back to crisis negotiation. I think this is just such a gorgeous example whereby the negotiators who are arriving on the scene and have to talk to someone who may or may not be in a suicidal crisis, they have to keep taking turns with whoever it is. They can't go home.
Elizabeth Stokoe:
They can't go give up. They have to keep going. One of the things they have to do is get the person to actually talk to them in the first place because we know that if you can get someone to take turns, then every single time the person actually takes a turn in the encounter, then they're choosing life every single time they take a turn. Even if it's a bit like, “Oh, I'm not doing what you're saying,” it's still not jumping or something. It's a really, really important environment to understand the importance of conversation analysis, I think. We know that quite often the negotiator will basically say, “Can we talk about how you are?” Or they'll ask to talk to the person. I think the NYPD crisis negotiation team's motto is something like talk to me. The reason why I think this is really important is that if you're going to build a motto and build training, then probably you're going to train your negotiators to ask the person in crisis to talk to you.
Elizabeth Stokoe:
When we look at the actual empirical examples though, the reality of these negotiations, we find that being asked to talk is very easy to resist, so people will say things like, “What's the point in talking? Talking doesn't do anything. I don't want to talk.” They almost respond to it ironically Like, “What are you asking me this for?” Whereas the negotiators … This is where the value of my work comes in, I think, which is I'm able to show you, and then back to negotiators, what it is that they are doing to nevertheless, have another crack at that because they have to. they can't just give up at this point. We see that negotiators will either … Some of them just don't ask to talk at all. They say, “Can we speak? I want to speak to you about X.” Those moments don't get resistance.
Elizabeth Stokoe:
People can't say, “Actions speak louder than speak” or, “I don't want to speak.” They just don't say that. They don't resist the verb speak and the broader action that that comprises, in the same way that there is this talk. The opposite of all of this is that if this was obvious, then no one would ask, “Can we talk?” They know already by remembering that speak better. It's more effective, but they and so what we do I can do is nevertheless, reveal the expertise of the negotiators who are managing something really effective and turn that into the training. Now we know what to train novice negotiators to say in terms of how they initiate dialogue with the person in crisis. I think the other thing to say at this point is that there's a tendency, I think, to not believe that language is that important.
Elizabeth Stokoe:
At the same time, it's fascinating, really. People think communication is really important and everyone will express that right up until the moment they want to scrutinize it properly and then develop guidance, training, and so on, from the ground, up, if you like. We tend to think that a person in crisis is going to either jump or not, on the basis of their psychology, not on the basis of what the negotiators say. Whilst their individual psychology is probably very interesting and relevant, we don't know what it is. It's back to where we started. If we don't know it, then negotiators just have to take turns anyway.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
Then we can actually find, when we bother to look, that the language is making a difference and now we know it.
Daniel Stillman:
It feels like, given the complexity of the situation, just giving people a Word document, a PDF, a set of pointers like, “Keep the person talking. Ask lots of questions.” The kind of advice that people crave and that we think is actually helping them may not actually help them become better negotiators. That's where I guess role-play comes into your training as well. I'm guessing.
Elizabeth Stokoe:
Yes. I have a view of role-play, which is when I started to do research-based training on the basis of some of these things that I was finding in the different environments that I was working, I became quite interested in, what was the alternative too? My way of taking this research back to practitioners is to show them real examples, real interaction in the wild, anonymized, and have them live through it, turn by turn. Then think about what they might do in any given moment, so see what a person in crisis or a patient or a potential client says at this moment in the encounter and then think about what they might do next in a real situation because I'm showing them real situations. Then we see what the practitioner did next and evaluate it. This is very different, I think, to role-play where you get an actor to play the part of the interlocutor. I became really interested in role-play as a thing and started to research on role-play.
Elizabeth Stokoe:
I was quite surprised at how widespread role-play is used as a completely uncriticized proxy for real interaction, first off. Or that where research is conducted as to the authenticity of role-play, it's basically a post-op, "Did it feel authentic?" kind of research question, which for me, is meaningless. People can say what they like, but it doesn't tell me anything as a conversation analyst. I basically conducted some research and started to compare in the wild. I started with police interviews with suspects and then looked at police interviews with actors playing the part of suspects, to see whether or not they look similar.
Daniel Stillman:
Surprisingly, they're different. I enjoy that.
Elizabeth Stokoe:
Surprisingly, they're different. Of course, role-play is one thing for maybe practicing strategies or something like that, but the issue is that sometimes it's the actor playing the part of the interlocutor, where they've been given bad information about the kinds of authentic things that a real patient, suspect, client, et cetera, might do. They've got a bad script, but the other thing, of course, and the police is a really nice example, the actor playing the part of a suspect isn't going to go to prison at the end of this encounter.
Daniel Stillman:
No.
Elizabeth Stokoe:
They can do anything, actually. They don't have the same stake in it. That stake is there, maybe to do a really good performance for the scenario, but they don't have that same interest that a real suspect has. We found differences and I think this is particularly consequential. If somebody's performance is being assessed and their job promotion, passing a course, depends on their performance in a simulation, that's where it becomes problematic, especially problematic, I think.
Daniel Stillman:
Sorry for interrupting, but in your book, I really enjoyed the fine slicing of secret shoppers calling veterinarians. Just to give people a sense of what you're noticing, that most of us don't miss, is where the ums are in a secret shopper. They don't actually know the breed of their dog. They pause at it. Whereas a normal person says, “Oh, I'm calling for this and I need to make an appointment about this.” They pause at a much later point. Just the placement of a pause or a word in the response of the person who is on the other side of the call can shift the conversation but the simulated call isn't a real call. The secret shopper is not acting the way a real person would. We're noticing those pauses maybe in subtle ways and reacting to them in uncomfortable ways.
Elizabeth Stokoe:
I think the other study that I've done, which you mentioned, is really nice because in the police study, obviously both parties in the simulation knew that they were in a simulation. Whereas in the secret shopper study, the person taking the call, which in this particular case was a vet practice, doesn't know that this is a secret shopper phoning to test out the experience that they have in the call. The mystery shopper is also going back, if you like, and reporting on the experience that they have in this call. My question is, do they have a basis for making any assessment? I guess the most extreme polemical version of my answer to that would be no, they have no basis for saying anything about the experience because for a start, they don't ask for the same things.
Elizabeth Stokoe:
They don't phone up and you can hear them sobbing already. I think this is one of the nicest examples, actually, where the mystery shopper is phoning. The remit is something like, “Show that you're upset about your animal and you're going to be asking about putting your animal to sleep.” Looking at how they do that, compared to how somebody who is genuinely phoning about a sick or dying animal, how they do that. The whole thing unfolds completely differently. Now, of course, for the person taking the call, they may or may not think something is off in the way this person that they're talking to is interacting on the phone.
Elizabeth Stokoe:
In a way, the mystery shopper whole thing, it relies on the fact, at least in terms of these phone calls, that you can't see anything about the person that you're talking to either. Mystery shopping works because when you are at the vet reception or in any organization, you don't generally think at any moment, I could be dealing with someone who is not a genuine customer or genuine client. It works in that way. I don't think the vet receptionist necessarily thinks that was a mystery shopper, but they might think they sounded strange because we are always saying things like, “They sounded strange.”
Daniel Stillman:
Yes.
Elizabeth Stokoe:
I think this is especially nice and gets us into conversational agents and Artificial Intelligence, and so on. Again, with a colleague, Saul Albert, and another colleague, William Housley, we've been looking at … You might remember that there was this small release of Google Duplex calls, this technology that could phone up and book a restaurant.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
Or make a hair appointment. Of course, we've been doing some work on that and looking at things like the placement of the ums and looking at, in very fine grain detail, how Google Duplex passes for humans.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
Of course, a massive part of that is that they're on the phone and that the human call taker is massively accommodating for the person that they think they're talking to. If you're on the phone with someone and someone sounds like they don't understand or they sound a bit off, then we don't think it's an AI. We just think they sounded really strange.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
We make up for it.
Daniel Stillman:
We try to repair it.
Elizabeth Stokoe:
Yeah. I don't know if you know, again, some conversation analysts conducted a study of all of these recordings of Lenny the chatbot that is online. I think if I understand this correctly, Lenny is something that you could download and plug into your telephone at home to deal with cold callers.
Daniel Stillman:
Oh yes. It befuddles them. Yes. I remember this.
Elizabeth Stokoe:
I think it's not AI. It's basically 16 possible things that Lenny could say in rotation. The challenge was how long could this elderly-voice sounding guy keep a cold caller on the phone? What you see is that the cold call just keeps accommodating who they assume they're talking to. They assume that an older guy, maybe he's confused or maybe just wants to chat. You can see that that when one party has no idea they're in any kind of simulation or AI or mystery shopper, they make it all happen.
Daniel Stillman:
Yes. Just to roll it back and to tie a bow on that, I just appreciate that your perspective is not uncritical when it comes to simulation and that we should question like any scientist would. Our observation of the experiment changes things. Our observation of the phenomenon changes things. I appreciate that as a scientist myself too. Our time flows away.
Elizabeth Stokoe:
Yes.
Daniel Stillman:
There's two big topics we haven't addressed. I don't know if we can do justice to both of them. I hope this isn't our last conversation. One of them is group conversations because as a facilitator, that's my main wheelhouse. A lot of people who are listening to this show are also leading meetings and changing organizations. These are much bigger, squishier, more complex conversations and it's harder to get data about them. I have a friend who actually tried to use subvocalization recordings to do mass measurements inside of corporations, but not surprisingly, people were resistant to the harnessing of that kind of data. We could say something about an organization by summing up and doing analysis on all of their conversations.
Elizabeth Stokoe:
Absolutely, especially-
Daniel Stillman:
Especially? Yes. I'm sorry.
Elizabeth Stokoe:
Meetings maybe will be a starting point.
Daniel Stillman:
Yes, exactly. That's the molecule of the conversations inside of an organization. What can you tell us about the application of your work to these types of group dialogues?
Elizabeth Stokoe:
I haven't really done research on big groups interacting. I think one of the difficulties is quite simply data capture, in the first place. If you just take a meeting as an example, if you imagine that you're in a room with 10 people, as a conversation analyst and you have all the resources of embodied conduct, gays, the material environment, and all of those things that are crucial to how we interact, then it's hard to capture enough that you can see what everyone is using to interact, all of the data that you can see that people are using to take turns and so on.
Elizabeth Stokoe:
It's not impossible, of course, but there is some research on meetings and so on. I'm mostly looking at stuff like turn allocation and how to get the floor, and who dominates the floor, those kinds of things, but I think it is a challenge for the analyst to get, if you like, all of … One of the reasons phone calls are so nice is because you are in the same position pretty much as those speakers you can't see.
Daniel Stillman:
Yes. It's stripped down.
Elizabeth Stokoe:
At least you've got the complete picture. I actually think it would be quite nice to look at these teams in Zoom meetings where you actually have a better sense of everyone's access because you can see the whole screen.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
You can see probably what most of the people can see. Actually, there's definitely an ideal time right now, while we're locked down, to look at how meetings are working online. I think there is also an issue around … The closest I've got to thinking about something like groups is to look at mediation. I've done a little bit of work on mediation, where there might be five people in the room. I'm particularly looking at, of course, how the mediators are facilitating that interaction. Of course, their job very much here is to ensure a balance of participation and so on. That's really interesting because as a model for facilitating any conversation, if your starting point is, we must get both perspectives out into the room equally, that's very different from many of the meetings where it could be like, “We want as few people as possible to talk.”
Elizabeth Stokoe:
Think about, what are you trying to do in a group? Are you trying to maximize participation or minimize participation? Because people might quite legitimately have different goals. Then at least one of the most interesting things in the mediation stuff is then looking at how the impartial perspective of the mediator and the process, which is meant to be impartial, can fall away within the first 30 seconds as the mediator opens up the encounter. For example, I've got a nice case where there are two or three parties who are in dispute and two mediators...One mediator is setting up. “This is what's going to happen.”
Elizabeth Stokoe:
“We're going to …” They're laying the ground rules and they say to the three parties who were all at war with each other, “Is it okay if we go by first names?” One person straight away, “Yeah. I'm fine with that.” Then the other mediator at this point jumps in and then checks in very explicitly in that socializing way that we were talking about earlier. Then checks in with both people who didn't immediately say yes. This one guy tends to jump in all the time. We all know this person is-
Daniel Stillman:
First movers.
Elizabeth Stokoe:
You ask the question and one person is right straight away. The challenge then for the person and doing any kind of facilitation is to decide, do I want that person to always jump in and be unchecked? Or am I going to very explicitly get an endorsement or an opinion from every other person in this room to stop that happening? Basically, what happens in this mediation is that they do that a little bit, but then you get the sense that it's becoming a bit awkward for them and so every time they say, “This happens,” then one participant will jump in and say, “Yeah. I get that.” They stop checking in with the other two in that rather awkward, passing the baton, type of way. “Do you understand? Do you understand?” They stop doing that. Before the mediation has even started, the participants have learned ... The guy who jumps in all the time, he's learned, I can say what I like and no one's going to check me.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
The other few people have learned he's always going to jump in and no one's going to check in with me.
Daniel Stillman:
There's a new racetrack being built. That's just-
Elizabeth Stokoe:
The racetrack is already completely skewed. You can see the effects of that all the way through. Really, I think setting up expectations of the kind of interaction that you're in, these opening moments are really crucial to understanding how else the whole interaction is going to unfold.
Daniel Stillman:
It's interesting because the facts that you shared way, way earlier, just when we think about the material of conversation, that we speak for two seconds and there's this 200-millisecond gap, and that longer gaps are viewed as awkward or rude. It's not surprising that group conversations can collapse into agreement very quickly and this maybe can transition us to touching on the last, maybe most important topic that we should do a whole other thing on, Ned who just seems like a terrible person. There's this conversation in your book where gender does play a role in group conversations. This group is having a very quick conversation about who's going to be the note taker and somebody was like, “Oh, well, my handwriting is terrible.”
Daniel Stillman:
The first one was like, “Oh, we need someone to be a secretary.” “Oh, my handwriting is terrible.” Ned is like, “She wants to do it. Secretary and female.” He collapses and this woman just takes the role on. She's like, “I'll just do that emotional labor.” She didn't say, “Well, who else has good handwriting?” Or, “Let's all write down everything.” These group conversations do have this way of collapsing very quickly, with movers and followers, with no opposers, but gender also really does shape conversations, it seems, but it's really hard.
Daniel Stillman:
I feel bad bringing this up where we are in the conversation right now because I think everyone wants to have the answer. What's the difference between men and women in conversations? You, as a scientist, will say, “Well, how do people come to have beliefs about sex differences in speech style?” How our beliefs encoded and enacted in self-presentation. It's not just like, “Oh, men are from Mars and women are from Venus.” We're like, “Well, how do we decide to act as a man?” What does it mean? How do we even build that racetrack? Then maybe how do we deconstruct that racetrack?
Elizabeth Stokoe:
My take on this goes back to things like men are from Mars, women are from Venus because that was the context in which I was doing my PhD. My initial project was going to be gender differences in things like turn allocation and who talks most, and who interrupts whom, and that stuff that was very prevalent in the early 90s, in social linguistics, actually. Showing that men dominate mixed-sex interaction and those kinds of things. Also, I wasn't ready to write a thesis that said, “No, no, no. Everything is equal now” because I live in the world. How many years later is it? Of course, we know that maybe things have even gone a bit backwards since the 90s, but in terms of what I wanted to do was that I wanted to try and move away from … Starting with the idea that there would be sex differences because things like men are from Mars, women are from Venus are actually which develop, which promote stereotypes as much as they do anything else and are really problematic, I think, in that way.
Elizabeth Stokoe:
Also, because there is, of course, since before then and since then, just such a massive, massive literature on gender performance, construction, and so on. Taking the lead from conversation analysis, I decided to look for moments where gender is demonstrably relevant to the people in the conversation and, of course, nominating somebody doing an action. How do you solve the problem of who's going to write down the notes? Well, one way to not do it, be the person asking, “Who's going to do it?”
Daniel Stillman:
That is a good piece of advice, by the way.
Elizabeth Stokoe:
“Who's going to take the notes?” “Well, secretary and a female.” My whole thesis, if you feel like, was built on observing this one moment in an interaction where you could see that it didn't matter whether there was a male style or a female style. In terms of asking the question, who's going to take the notes, or is someone writing? There's nothing gendered about saying, “Is someone writing?” Men don't do that more than women, any of the gender stuff to think about. When you start to take those actions again like we talked about right from the start, then a lot of these apparent differences just really fall away empirically, even if we believe them, a bit like the body language stuff, but that doesn't mean that you can't nevertheless show that gender or any other characteristic or category is creeping into the interaction, creeping back from the interaction, being pulled into, being pushed out of, and having an impact on the unfolding organization of the racetrack.
Elizabeth Stokoe:
That particular case was my case in point and what I like about that is that it doesn't rely on me saying anything at all about gendered styles, which I really wanted to get away from because what I was able to show was here are moments of consequence, where gender is being recruited into the conversation to do something. Then basically, the challenge was to try and go from one case and think, how on earth am I ever going to find that again? To over then the next decade or more, building up data sets where I could show systematic things happening with categories. A really simple example, the most recent example of that that I thought I would just point out because I think it's a nice one is I started to find things like when people are denying certain sorts of things that they might have been accused of, they might invoke a category in response and that might be gendered.
Elizabeth Stokoe:
I then started to look at how when people are phoning up to buy things, in a particular case, it was buying windows, how the salesperson does a really straightforward thing, and becomes gendered. The straightforward thing is, ask for the customer's name and they tend to do it in one of three ways. They'd either say, “Can I take your surname?” Or, “What's your name?” Basic questions. Or they might say, “So you're Mr.?” You'd say, “Stillman.” That latter method, “So you're Mr,” waiting for you to finish off, was only used with men.
Elizabeth Stokoe:
Women were not invited to complete the turn in that way, so people were not asked, “So you're Mrs.?” Instead, they were asked, “So is it Miss, Mrs., or Ms.?” Which sometimes led to problems. Then the other thing, basically, what this shows overall, of course, is that even in the late part of the 20, teens or something, that we still have this kind of standards address term for men. All men can respond-
Daniel Stillman:
Yes.
Elizabeth Stokoe:
To the all-men, gender, sexual orientation, all of those things. It doesn't matter, men. Whereas women's titles are still a hugely tricky thing in interaction.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
We have a couple of moments where if a woman referred to herself as Ms, there was a little laugh from a salesperson. You can feel this language after all. Of course, all of that is also wrapped around the idea that the person you're talking to is straightforwardly categorizable as a man or a woman.
Daniel Stillman:
Yes.
Elizabeth Stokoe:
Probably straight, all sorts of assumptions that are built into just the most simplest of actions, which is asking somebody their name so that you can sell them some windows. There's a lot of work still to be done, I think.
Daniel Stillman:
Well, I mean, there's a lot more to talk about, but we're out of our time.
Elizabeth Stokoe:
Yes.
Daniel Stillman:
Liz, I really appreciate you talking about this. Obviously, talk, I think our lives are built of it and it's super important for us to understand it and so I really appreciate your contribution to our understanding of it.
Elizabeth Stokoe:
It's been a pleasure to talk to you.
Daniel Stillman:
Thank you. People should go buy your book. Is there anything else? People should find you on the Internet where?
Elizabeth Stokoe:
If you Google my name, you'll find all sorts of books online and so on. I hope that people will just open their minds a little bit to what you can find out if you study in this very fine-grained way that we do conversations.
Daniel Stillman:
Thank you so much, Liz. I really appreciate it.
Elizabeth Stokoe:
It's been good to talk.
Daniel Stillman:
Thank you so much