We imagine surveys, diagnostic instruments, police interviews and exam questions are asked identically every time because there's a script. In reality, people cannot help embellishing, tweaking and adding to questions. There is an interactional imperative that makes it very hard to sound like a form, even when the law or protocol says to deliver it verbatim.

Standardized instruments are rarely delivered as written
Open vs closed questions is itself partly a myth ('do you have Wi-Fi?' isn't yes/no)
Teachers, psychiatrists and police all add embellishment to scripted questions
Multi-part questions leave the respondent guessing which bit to answer

“Even if you're a police officer in the situation where the law says say it like this, it doesn't quite look like that”
Elizabeth Stokoe · 30:00

#scripts#questions#conversation-analysis#assessment

Myth Buster30:00

Ums and ahs aren't errors, they're doing precise work

Filler sounds are treated as mistakes to erase, but um, ah, you know and okay are not interchangeable and each does specific work: signalling thinking, word-searching, difficulty, or delicacy. Comparing real vet callers with mystery shoppers shows the ums land in different places, evidence that they aren't random noise but structured parts of talk.

Um, ah, you know and okay each do different interactional jobs
Real transcripts of anyone, including politicians, are strikingly messy
Mystery shoppers say 'a dog'; real owners say 'my dog', and their ums sit differently
Judging people as less intelligent for filler words is stereotype, not evidence

“the ums and uhs aren't randomly splattered everywhere. They are, when you look at enough of them, they're doing particular bits of work in particular…”
Elizabeth Stokoe · 33:00

#filler-words#speech#conversation-analysis#myths

Myth Buster38:00

The 93% nonverbal communication myth, busted

The famous pie chart claiming communication is 93% nonverbal and 7% verbal collapses on contact: radio, talking in the dark, and phone calls would be impossible if it were true. Even Albert Mehrabian, whose studies spawned the figure, has disowned the way 'communication consultant gurus' slap it on pie charts; his actual finding was narrower and more complicated.

If 93% were nonverbal, radio and phone calls couldn't work
The claim is compelling because people love a clean statistic and a pie chart
Mehrabian himself is irritated by the myth built on his work
Body-language lie-detection cues have little evidence and are rarely studied in the wild

“if I now turned off the audio, this podcast would make sense still for 93% of the time. We immediately know that can't be true”
Elizabeth Stokoe · 38:30

#communication-myths#nonverbal#body-language#myths

Explainer· 3

Explainer00:30

Why 'Hi, how are you?' tells you everything about a conversation

The seemingly pointless 'Hi, how are you? Fine, how are you?' opening is a diagnostic. Its presence signals a routine, low-stakes exchange; its absence signals urgency, emergency or imminent conflict. Contrasting openings across thousands of real conversations reveals that this filler talk is doing precise interactional work.

The 'how are you' exchange marks a mundane, non-urgent, conflict-free start
Skipping it and jumping straight in ('Chris?' 'Liz?' 'What's the deal?') flags a coming argument or emergency
Cold callers whose voice you don't recognise trigger guardedness from the very first seconds
The opening seconds of a conversation already leak its emotional stakes

“that apparently pointless filler talk is actually telling you a lot about the kind of interaction this is that's about to happen”
Elizabeth Stokoe · 01:00

#conversation-analysis#openings#small-talk#communication

Explainer14:00

The 'first mover' problem and breaking the fourth wall

A 'first mover' takes a problematic opening turn ('I don't really like that jumper') and, if challenged, flips it so the offended person becomes the problem ('I was just joking'). Calling it out breaks a social convention, so it feels like a record-scratch, especially in a group, even though the original jab did not. People let a lot slide precisely because challenging is so hard.

A snide opening turn traps you: object and you become 'the one who can't take a joke'
The convention protects the offender, not the offended
Breaking the fourth wall to name it feels jarring to everyone present
We let things slide to avoid being angry all the time

“somehow if you call them out on it, you're the problem by being offended rather than the person actually being offensive in the first instance”
Elizabeth Stokoe · 14:30

#conflict#social-dynamics#conversation-analysis#boundaries

Explainer20:30

The one-second rule: why long silences feel like nails on a chalkboard

In natural talk, gaps run 100-200 milliseconds; a full second is already a marked delay, and 2-3 seconds is huge. Elon Musk's 31-second pause on Lex Fridman and a 16-second Lady Gaga silence (cut before broadcast) are extreme outliers. Long silences feel strange because, empirically, they almost never happen, and we read every gap as a signal that something needs repair.

Normal conversational gaps are 100-200ms; one second is noticeably long
The longest silence in Stokoe's data is 11.2s, in a police suspect interview
TV cut Lady Gaga's 16-second silence because viewers couldn't bear it
A gap triggers 'repair': you assume your question was unclear and redesign it

“you're looking at about 100 to 200 milliseconds... a silence of 1.0, like a complete a whole second, is starting to be quite a delay”
Elizabeth Stokoe · 21:00

#silence#timing#conversation-analysis#podcasting

Story· 3

Story01:30

How people in danger call 999 by 'ordering a pizza'

When a caller is in danger but cannot speak freely because the threat is in the room, they use the normal furniture of a friendly call to smuggle a signal. Doing 'how are you' small talk to a 999 dispatcher, or saying 'I'd like to order a pizza for delivery', lets a trained call-taker hear a genuine cry for help without the caller ever saying the dangerous words.

Callers reproduce friend-call rituals so a nearby threat hears a harmless conversation
Dispatchers learn to read the mismatch and hear a real emergency in coded phrasing
Police call-takers must separate these from genuine nuisance calls, and do so quickly
The signal works because the ritual is out of place for a 999 call

“People say things like I'd like to order a pizza for delivery. How does the call taker hear that as a genuine request for help”
Elizabeth Stokoe · 02:30

#emergency-calls#covert-communication#conversation-analysis

Story05:00

Lenny the chatbot: 16 phrases that trap cold callers for hours

Lenny is a bot built to keep cold callers on the line, made of only about 16 pre-recorded, shuffled phrases in the voice of an elderly Australian man. With no real AI, it convinces callers they are speaking to a confused old gentleman, and they feel too polite to hang up. Recordings of Lenny's marathon calls circulate on YouTube.

No algorithm, just ~16 shuffled recorded phrases and a convincing persona
Callers stay on because leaving feels rude to the apparent old man
Illustrates how much conversational cooperation runs on social convention, not content
Sits in the same online-satisfaction bucket as scam-baiting channels

“it was just something like 16 phrases pre-recorded and then shuffled, and it sounded like the voice of an elderly Australian man”
Elizabeth Stokoe · 05:30

#cold-calling#chatbots#social-convention#tool

Story64:30

The 'that's right' moment that ended a hostage crisis

Chris Williamson recounts Chris Voss's technique of earning a 'that's right' by steel-manning the other person's position so precisely they feel fully understood. In a stalled overseas hostage negotiation, Voss's understudy delivered a long, accurate account of the hostage-taker's grievance and got 'that's right'; within 24 hours all hostages were released, though the captor fled.

'That's right' comes from showing you understand exactly why someone feels as they do
It differs from 'you're right', which people say just to end a conversation
The captor released everyone within 24 hours after conceding 'that's right'
Stokoe warns empathy misfires badly if delivered too early, before enough evidence

“the that's right is about steel manning the other person's position. It's about showing them that you understand precisely why they feel the way that…”
Chris Williamson · 65:00

#negotiation#empathy#steel-manning#story

Takeaway· 4

Takeaway08:00

The Wi-Fi code test: give people what they need before they ask

Two real cafe field notes contrast two service styles. In one, asking 'do you have Wi-Fi?' gets a literal 'yes' and the customer has to drag out the code. In the other, staff volunteer 'if you need the Wi-Fi, here's the code' as the tea arrives. Anticipating the real need, briefly and unprompted, makes people feel warm toward you and cuts pointless back-and-forth.

'Do you have Wi-Fi?' is a request for the code, not a yes/no question
Answering literally forces the customer to keep pushing and feel like a burden
Volunteering the need met, unbidden, builds instant goodwill
It works precisely because it is short and skips scripted rapport-building

“as your tea comes across the desk the cafe staff member says oh by the way if you need the Wi-Fi here's the code”
Elizabeth Stokoe · 09:00

#service#efficiency#rapport#conversation-analysis

Takeaway11:00

The vet receptionist test: read whether they want rapport or the price

A caller asks a vet how much puppy injections cost. A one-word reply after a 0.7-second delay ('Victor') signals they just want the price, yet the receptionist ploughs on with 'is it your first puppy?'. Being nimble enough to hear who wants to chat about their pet versus who wants information, and matching them, is the core skill, and it is hard to train.

A short, delayed answer signals someone wants information, not conversation
In conversation analysis a 0.7-second pause is already a large delay
Some callers are dying to talk about their puppy; some just want the number
The skill is reading the cue in real time and adjusting, which not everyone can do

“can you be nimble enough and listen enough to actually think, right, this is someone to do that with cuz they're dying to do it.…”
Elizabeth Stokoe · 12:00

#listening#rapport#service#conversation-analysis

Takeaway16:30

Why not saying 'thank you' first feels physically painful

Service calls end in a fixed ritual: 'thanks, thanks, bye-bye', with the service user obliged to close first. When a receptionist tries to end before the user has what they need, the user must jump back in to push for it. Stokoe's challenge: next time you're unhappy on a call, try withholding your thank-you, and notice how horrible it feels, which shows how strong the convention is.

Ending a service call is the user's job, marked by 'thanks, thanks, bye-bye'
A provider ending early forces the user to reopen and push past them
Withholding your thank-you feels genuinely unpleasant, revealing the norm's grip
The exercise exposes how much conversational ritual controls us

“the next time you're unhappy on a telephone call, try not saying thank you at the end... You'll feel really horrible”
Elizabeth Stokoe · 17:30

#politeness#phone-calls#conversation-analysis#social-norms

Takeaway26:00

Stop offering a menu of answers after you ask a question

Nervous interviewers ask a question then immediately tack on options ('was it as a kid, or was it...'). This is driven by discomfort with silence, but it backfires: an open question already invites any answer, and adding options forces the guest to pick one, reject them, or break the frame you've imposed. Better to ask, stop, and let the question sit.

Menus of options come from anxiety, not from helping the guest
Guests rarely say 'it's neither' and instead take a suggested answer
Options constrain the guest into the frame you've set
The fix is 'question, stop, silence' and trusting the question

“it's mostly about the comfort of question, stop, silence”
Chris Williamson · 27:30

#interviewing#questions#podcasting#conversation-analysis

Frameworks in this episode

Explain-as-Process, Not Ethos

Resistance-Reducing Verb Choice

Insights & moments

⚡Myth Buster· 3

The myth of the standardized script: even police interviews get embellished