Do I actually listen to people, or am I just waiting to speak?

You do both at once, which is why the answer feels like a verdict. Drafting and listening draw on the same narrow resource, and conversation forces you to run them together, so the reply starts building before the speaker stops. You hear the words. The meaning is what gets spent.

If you have ever tried it, you know the shape of the workload: keep what you want to say at the front of your mind, and process everything new the other person says, at the same time. That is not a manners problem. That is a workload, and it has been measured.

In an EEG study, Sara Bögels and colleagues found production planning starts within half a second of the moment the answer becomes available, with an attention switch from comprehension to production in the same window.

What does holding a draft cost my understanding?

It costs accuracy on the incoming language, not on sound in general. When people plan a reply while a question is still arriving, comprehension of that question gets worse, and the interference is specific to linguistic processing. You can be wide awake, focused, and still lose the point to your own sentence.

The trade is not a metaphor. In task-oriented dialogue, Mathias Barthel and colleagues found utterance planning begins before the incoming turn ends, and described the early start as costly while being what makes turn timing efficient. Both halves are true, which is why the fix cannot be "plan later."

A 2026 result sharpens it: phonological input processing is reduced during speech planning in turn taking. The reduction is in processing the incoming speech, not in hearing it. Your ears work. Your parsing does not.

Why does my brain start drafting before they stop talking?

Because the gap between turns is shorter than the time it takes to build a sentence, so the only way to answer on time is to start before the other person finishes. Early drafting is not a bad habit. It is the mechanism that makes conversation fluent, and it has been measured in real conversations, not only in labs.

Planning begins before the incoming turn ends in scripted task dialogue, and Bögels found the same early start in free interviews. A 2026 preprint released a dual-EEG hyperscanning dataset of 18 dyads in natural French conversation, built to study turn exchange and response preparation.

None of this is your fault. Without early planning, conversation would stall and everyone would sound like they were thinking mid-sentence. The system is doing its job. The job just has a bill, and it lands on comprehension. The turn-taking clock itself is a separate story.

Why do I lose my own point while I wait for them to finish?

Because a held sentence is a live job, not a stored file. Keeping it available means rehearsing it, and rehearsal competes with the incoming speech for the same resource. The longer you wait politely, the more the draft decays, which is why the point often returns in the car on the way home, fully formed and completely useless.

This is the cost nobody warns you about. The comprehension you spent on drafting is gone, and the draft is not safe either. The 2015 EEG paper notes that comprehension may suffer once production planning starts, because both processes draw on largely the same cognitive and neural resources. Holding a sentence is not free storage. It is a second job.

So the trade is symmetrical. Hold the sentence and you lose the meaning. Drop it and you lose your turn. Trying harder is more of the same resource.

How do I know when someone has actually finished talking?

Poorly, and for a structural reason. A pause and an ending look identical from the inside, and the cues that mark a real ending act as launch signals for a reply that is already drafted. That is why people trying hard to be polite still interrupt: they detect the end late rather than deciding to cut in.

Barthel, Meyer and Levinson separated preparation from initiation: next speakers plan their turn early and speak after turn-final go-signals. Corps and colleagues found listeners use content predictions to determine what to say but not when to say it. Your system knows what it wants to say, not when to say it, and it treats the first plausible cue as permission.

The drill is perceptual, not polite. A pause is not an ending. Listen for the signal that the thought is complete rather than the signal that the breath ran out.

How do I listen without holding a draft?

By not holding one. The trade only runs while a full sentence is open, so the move is to let the sentence go and keep the intention. You will feel underprepared, and that feeling is the cost of the trade being switched off. What you get back is the part you were spending: the actual point.

The fix is not a listening technique. It is a production decision. Retrieval practice works because producing the answer is what stores it, as Roediger and Karpicke showed in Psychological Science in 2006, and holding a draft hands production to a sentence that has not been said yet.

If you want to train that specific muscle, see Absorb on the App Store and judge it by what you actually absorbed. Fluency is subsidised. You can keep paying in comprehension, or pay in polish and keep the point. The second is cheaper than it feels.