Modern WisdomThe Science Of Analysing Conversations - Elizabeth Stokoe
CHAPTERS
- 0:00 – 1:42
Small talk that signals safety: why “How are you?” isn’t meaningless
Stokoe explains that “Hi, how are you?” looks like empty small talk, but it does important interactional work: it signals normality, lack of urgency, and low conflict. By comparing openings across settings, she shows how skipping or reshaping this sequence can instantly reframe what kind of encounter is about to occur.
- •“How are you?” is a routine opener that establishes a non-urgent, non-conflict frame
- •Removing it can signal urgency or impending argument
- •Openings reveal the “type” of interaction within seconds
- •People respond differently when they recognize a familiar vs unfamiliar voice
- •Conversation analysis focuses on real-time unfolding interaction, not recollection
- 1:42 – 3:19
Covert emergencies and call-taker inference: signaling danger without saying it
The discussion moves to emergency calls where callers must communicate that they’re in danger while someone nearby cannot know they’ve called for help. Stokoe describes how callers use ordinary conversational conventions (including friendly openers or cover stories like ordering pizza) and how dispatchers learn to detect these cues amid nuisance calls.
- •Callers may need to disguise the request for help from someone present
- •Routine conversational structures can be used as cover (e.g., friendly openings)
- •Examples like “ordering a pizza” function as indirect help-seeking
- •Dispatchers must distinguish genuine covert requests from prank/nuisance calls
- •Understanding comes from how sequences unfold turn-by-turn
- 3:19 – 6:51
Cold calls, chatbots, and the moment we detect “this isn’t human”
Chris and Elizabeth explore sales-call openings and why “How are you?” can backfire when the recipient doesn’t recognize the voice. They discuss pre-recorded scripts and the ‘Lenny’ chatbot, illustrating how humans rapidly detect timing, pitch, and pacing that don’t fit natural conversation—and how politeness norms can still trap callers into continuing.
- •Unrecognized voices make recipients guarded; responses become clipped or silent
- •Bad sales scripts ignore cues that the other person wants to end small talk
- •Pre-recorded calls often fail because timing and prosody feel ‘off’
- •Lenny chatbot: simple phrase loops can keep cold callers engaged
- •Politeness and social accountability can prolong unwanted interactions
- 6:51 – 12:13
How language steers outcomes: interaction as the engine of everyday life
Stokoe answers how much humans are “pushed and pulled” by language, emphasizing that social life is accomplished through interaction—words plus embodied conduct (gaze, gesture). She uses service encounters to show how tiny design choices (what gets offered, when) create friction or warmth and dramatically change the experience.
- •Interaction is how we request, offer, invite, and get things done
- •Manipulation isn’t always sinister; it’s built into coordination
- •Service examples show how failing to infer the ‘real request’ creates friction
- •Proactive offering (e.g., giving Wi‑Fi code unprompted) builds rapport efficiently
- •Conversation analysis can pinpoint why an exchange ‘felt’ burdensome
- 12:13 – 19:41
Judging personality from talk: “first movers,” offense, and the difficulty of calling it out
They unpack how we infer traits like rudeness, warmth, or entitlement from what people do in conversation—often from the opening move. Stokoe discusses the social risk of challenging someone’s ‘snide’ first turn, norms around ending calls (the pain of withholding thanks), and micro-actions of allyship (including strategic silence).
- •We infer ‘personality’ rapidly from conversational behavior, not questionnaires
- •“First mover” openings can trap others into conflict immediately
- •Calling out unfairness can feel more disruptive than the original slight
- •Endings are structured; saying “thanks” is socially hard to withhold
- •Allyship can be tiny (a look) or direct; silence can prompt self-repair
- 19:41 – 26:03
Silence as action: timing, repair, and why long pauses hurt
Stokoe explains that silence isn’t inert—its length and placement are meaningful and often initiate repair (redesigning a question, revising an invitation). With examples from podcasts and police interviews, she shows that even one second can be a noticeable delay, while longer gaps become ‘breaches’ that trigger discomfort and interpretation work.
- •Conversation runs at ~100–200ms between turns; 1 second is already a delay
- •Long silences are rare in daily talk, so they feel empirically ‘strange’
- •Silence often prompts repair: speakers redesign questions or offers
- •Examples: invitation silences lead to add-ons (“Or Saturday?”)
- •Certain contexts demand rapid responses (“I love you,” appearance questions)
- 26:03 – 30:09
Why we add “menus” to questions—and how standardized scripts break in real life
Chris reflects on inexperienced interviewing patterns: asking a question and immediately adding options to avoid silence. Stokoe generalizes this to regulated settings (surveys, police interviews, oral exams), arguing that truly standardized delivery is difficult—people naturally embellish, reformat, or stack questions, which can distort responses and fairness.
- •Adding options constrains answers and leads respondents unintentionally
- •Standardized instruments rarely remain standardized in real interaction
- •Multi-part questions steer recipients toward the last component
- •Institutional talk has interactional pressures that resist ‘form-like’ delivery
- •Open vs closed questions are less clear-cut than people assume
- 30:09 – 38:14
Ums, ahs, and ‘like’: not errors, but tools for thinking, delicacy, and design
Stokoe reframes fillers as functional elements that help speakers manage word search, difficulty, and sensitive topics. She shares a study comparing real vet callers with mystery shoppers, showing that fillers aren’t randomly placed: their positioning can reveal authenticity and the interactional job the speaker is trying to do.
- •Ums/ahs can mark thinking, word search, trouble, or delicacy
- •Transcripts look ‘messy’ because real talk is messy
- •Fillers are systematically placed, not splattered randomly
- •Mystery shoppers vs real callers differ in pronouns (“a dog” vs “my dog”) and filler placement
- •Judgments about intelligence and competence are often stereotyped and contextual
- 38:14 – 41:27
Debunking communication myths: the ‘93% non-verbal’ claim and body-language certainty
Stokoe dismantles the popular statistic that communication is 93% non-verbal, showing how it collapses under basic counterexamples (radio, darkness, foreign languages). She acknowledges embodied communication matters, but criticizes simplistic body-language rules (looking right = lying) due to lack of naturalistic evidence—highlighting conversation analysis’ ‘in the wild’ approach.
- •The 93% non-verbal pie chart fails obvious real-world tests
- •Mehrabian’s work is widely misused; even he has challenged the myth
- •Embodied cues matter but don’t replace language in most interaction
- •Body-language ‘tells’ are rarely validated in natural settings
- •Conversation analysis prioritizes real, consequential interactions over lab simulations
- 41:27 – 46:19
Nonverbal coordination in real life: gaze, ‘doing nothing,’ and signaling non-threat
Using a crowded train story, Stokoe illustrates how a brief shared look can coordinate vigilance and solidarity without words. They discuss how people ‘do nothing’ in socially recognizable ways (waiting to be served vs waiting for someone), and how micro-displays communicate safety, neutrality, and non-threat in public space.
- •A single glance can create mutual understanding and coordination
- •People calibrate gaze finely: what counts as ‘a look’ is socially learned
- •‘Doing nothing’ is an interactional achievement with recognizable displays
- •Neutral friendliness (closed-mouth smile) can signal non-threat
- •Meaning depends on setting and shared expectations, not isolated gestures
- 46:19 – 53:43
First-date interaction mechanics: smooth topics, texting gaps, and visible self-editing
Stokoe draws from speed-dating research to outline what makes dates flow: staying mundane early, asking non-intrusive questions, and finding expandable topics. She contrasts spoken dating with messaging, where bursts of texts and sudden silences create anxiety, and where screen-recording reveals the hidden drafting/deleting that spoken conversation masks.
- •Early date talk works best when it avoids overly dramatic or scripted prompts
- •Successful interaction involves reading cues and adapting turn-by-turn
- •Messaging shows rapid exchanges followed by delays that feel consequential
- •Screen recordings reveal drafting/deleting and cautious self-presentation
- •Speed-dating data shows delicate questions (e.g., relationship history) produce repairs and reformulations
- 53:43 – 1:00:55
Hostage & suicide crisis negotiation: turning points, word choice, and getting traction
Stokoe describes analyzing police crisis negotiation recordings to identify turning points that move someone from danger toward safety. A key finding: proposals to “talk” attract resistance (“talk is cheap”), while “speak” gains faster traction; similarly, “help” can be resisted, whereas “sort things out” and action-oriented formulations progress the interaction.
- •Negotiations often succeed; the analytic goal is finding turning points
- •Early phase focuses on physical safety before deeper negotiation
- •“Talk” proposals often trigger resistance; “speak” proposals reduce it
- •Crisis subjects may reject generic care claims (“you don’t know me”)
- •Action-oriented language (“sort things out”) tends to gain cooperation
- 1:00:55 – 1:15:44
Reverse-engineering effective communication: from nudges to mediation ‘process’ talk
The conversation broadens to how small linguistic tweaks can systematically change outcomes (e.g., willingness vs interest; social-norm towel signs). Stokoe explains her applied work with mediators: describing mediation as a clear process (steps and sequence) is more persuasive than selling an ethos (impartiality), and training works best when it builds on practices already present.
- •Micro-choices in verbs and formats can tilt next-turn outcomes
- •‘Willing’ can secure agreement where ‘interested’ or ‘like’ may not (with caveats)
- •Nudge examples are more informative when you analyze the exact language used
- •Mediation uptake improves when described as a step-by-step process, not just values
- •Training is easier for scripted openings than moment-by-moment improvisation
- 1:15:44 – 1:16:15
Wrap-up: where to follow Elizabeth Stokoe
Chris closes by asking where listeners can find more of Stokoe’s work. She directs people to her Twitter handle, and the episode ends with standard show sign-off.
- •Stokoe shares her primary online location
- •Chris thanks the guest and audience
- •End-of-episode navigation to clips and subscription prompt
