You've been treating conversation as a thing you need permission to enter and an excuse to leave. It's a system with published rules, and both of those are in it.
This is the module you asked for after the first three: getting into a standing group, and getting out of a conversation you're stuck in. They look like two different problems — one is nerve, one is manners. They're the same problem. In both cases you're waiting for someone else to issue something the system says you issue yourself.
Everything in this lesson is documented mechanics. Conversation has a grammar; linguists wrote it down decades ago; and a large share of what you've been reading as your own social defect is simply not knowing rules that exist. That's a much better position to be in than having a defect, because rules can be read.
Across ten languages, the mean gap between one person finishing and the next starting is +208 milliseconds. The modal gap is 0–200ms. Every language tested clusters near zero. Stivers et al. 2009, PNAS 106(26)
Sit with that number for a second, because it dismantles a rule you've been running. Two hundred milliseconds is faster than you can decide anything. It is not enough time to notice a gap, evaluate whether it's really a gap, compose a contribution and deliver it. It is barely enough time to blink.
So the clear opening you're waiting for does not exist. If your rule is "wait for an unambiguous gap," you've adopted a rule that structurally guarantees you never speak. That isn't a failure of nerve on your part — it's a rule that cannot be satisfied, executed correctly. Which is precisely the experience you described in groups: everyone else seems to find a moment, and you never do.
The people who do speak aren't finding gaps. They're projecting the end of the current turn and launching into it as it arrives. That's a different operation entirely, and it's the one the next section licenses.
Turn-taking is formally specified. At each point where a turn could legitimately end — a transition-relevance place — three rules apply in order:
Read Rule 2 again. Self-selection isn't tolerated, it's specified. It is the system's own provision for what happens when no one has been nominated — which, in a loose social group, is nearly all the time. Sacks, Schegloff & Jefferson 1974, Language 50(4)
Your belief that contributing would be "interrupting the group" is a misreading of a system that explicitly authorises you. And the thing you're most afraid of — starting fractionally before someone has quite finished — has a name, terminal overlap, and it reads to everyone present as evidence you were tracking the turn correctly. It is not a breach. It's the sound of someone participating. Schegloff 2000
Now the other half, and the more urgent one for you.
You don't exit a conversation by declaring an exit. You exit by opening a pre-closing sequence — a small token that signals the topic could end here and offers the other person a turn to either release you or add one last thing. "Well…" "Okay…" "So…" Or a summarising line that rounds the topic off. If they take the release, you both produce a short terminal exchange and it's finished, and neither of you experiences it as abrupt. Schegloff & Sacks 1973, Semiotica 8(4)
The half-beat matters. It's the slot where they get to release you, and skipping it is what makes an exit feel like a walk-out. Give them the beat and the exit is collaborative — they closed it too.
This solves both of the trapped scenarios you described.
The monopoliser. They aren't holding you because you lack a good enough excuse. They're holding you because you were waiting for them to close, and closings have to be opened by somebody. They have no reason to open one — they're enjoying themselves. So it falls to you, and it is not rude to do it, because opening a pre-closing is an offer, not a rejection.
The person who turns to a third party. They've just handed you a free pre-closing. The turn has been formally reallocated; you are not required to stand there waiting to be reassigned one. "I'll let you two catch up" is a complete and entirely legitimate exit, and it costs nothing.
The last piece, and the one that converts guesswork into an observable signal.
People standing in conversation maintain a shared, protected empty space in the middle of the huddle — the o-space — which non-members are expected not to walk into. Entry to the group is granted when the group physically reconfigures: someone rotates their shoulders, or takes a half-step back, opening a gap in the perimeter. Kendon 1990, Conducting Interaction, Cambridge UP
That reconfiguration is the invitation, and it arrives before anyone speaks. So the move is: approach the perimeter, stand at a comfortable distance, wait a beat. If the circle opens, you're in — no verbal permission needed, and no opener required. If it doesn't open, that's your answer.
Note what that second case gives you. A circle that doesn't open is telling you something about the group's current state — a private conversation, a closed pair, a moment you've walked in on. It is not a verdict on you, and critically, you now have a signal where you previously had a guess. Most of the discomfort in approaching groups comes from not knowing whether you're welcome. The information was always being broadcast; you just weren't reading that channel.
Four exits using the formula. Pre-closing token, forward-looking reason, half-beat pause. At least one of them, every day, must be from a conversation that is going well. That's the rep that actually does the work.
Two group approaches where you read the o-space before deciding anything. Walk to the perimeter, wait a beat, and let the circle answer.
One deliberate self-selection in a group conversation — take a turn nobody handed you. Once a day. It only has to be one sentence.
Three days of exits. Tap as you go.
| Clean exits | Exits where you opened the pre-closing and left inside about twenty seconds. Target 4/day. |
| Good-conversation exits | What share of those exits came out of conversations you were enjoying? This is the number that matters. If it's near zero, you haven't learned to exit — you've learned a nicer way to escape. |
| Group entries attempted | Approaches made, not approaches accepted. The circle's answer isn't yours to control and isn't part of the score. |
The second row is the whole design. An exit skill that only fires when you're uncomfortable isn't a skill, it's the old avoidance with better manners on top — and it will keep you out of exactly the conversations that were working. Leaving something good, on purpose, while it's still good, is what proves the exit is under your control rather than your discomfort's.
Week 3 ended on a checkpoint worth carrying into this week: when I switched topics, was the thread genuinely exhausted — or was I uncomfortable and wanting somewhere else to be? The exit formula makes that question sharper, not softer. You now have a clean, polite, socially fluent way to leave any conversation at any moment. If your exit count is high and your conversation lengths are collapsing, the tool has been captured by the thing it was meant to fix.
The Barcelona Improv Group Sunday Open Workshop, 17:00–19:00, €10, capped at 18 and explicitly beginner-friendly. It's the right venue because group entry is not a side-effect there — it's the entire activity. Email classes@barcelonaimprovgroup.com first: it's confirmed as their standing drop-in, but I couldn't verify it runs through August.
Better-verified fallback: Barcelona Language Exchange runs six nights a week with no August break — Mon/Thu Space Cowboy (Born) 19:00, Tue Jardinet d'Aribau 19:00, Wed Soda Bus 19:00, Fri Estació de França 20:00, Sat Trafalgar Pizza Club 20:00. €4–5 minimum consumption. Rotating tables means exits are structurally expected, which makes it near-ideal for this particular drill.
From memory: give the exit formula in three parts, say what a closed circle tells you, and name the one number you're logging this week.
Exit: pre-closing token + forward-looking reason + half-beat pause. A closed circle tells you about the group's current state, not about you — it's information, and having it is the point. The number is the share of your exits that came out of conversations that were going well.
Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). "A Simplest Systematics for the Organization of Turn-Taking for Conversation." Language, 50(4), 696–735. DOI
The paper that established conversation has a formal structure at all. It's long and technical; the part you want is the rule-set in section 2, which is a couple of pages. If you'd rather read something short and free, Stivers et al. 2009 is open-access, five pages, and contains the 208ms finding — which on its own is enough to retire the rule you've been running.