When Conversation Flows

Reading guide

The teaching in one sentence

Talk goes somewhere exactly when at least one person in the room has hold of what is being discussed.

Kathā pavattinī

Kathā is talk, discussion, a discourse in the ordinary sense — the same word behind the 話 of a Chinese conversation and the name of the Abhidhamma’s Kathāvatthu, its book of disputed points. Pavattinī is from the verb for rolling on, proceeding, keeping going.

So the image is of talk that rolls rather than stalls. Sujato’s “flows” catches it. The discourse is not describing pleasant conversation or agreement; it is describing a discussion that has momentum, one that gets somewhere by continuing.

That is a recognizable thing, and the discourse asserts that it has exactly three conditions, all of them about comprehension and none of them about rapport, seniority, or good will.

What changes when a reason becomes a description

AN 3.43 gave these three as considerations that justify teaching — a prospective judgment, made before you speak. AN 3.44 gives them as the circumstances in which conversation in fact proceeds — a retrospective observation, available after.

Putting them together produces a diagnostic that neither discourse states on its own. If you had a good reason to teach and the conversation is not moving, one of the three conditions you thought was present is not. The reason and the result are made of the same material, so a failure in one is evidence about the other.

It also removes an excuse. A stalled discussion is not explained by the audience being unwilling, or the hour being late, or the subject being difficult; on this account it is explained by nobody in the room having hold of the meaning or the formulation. That is an unforgiving diagnosis and it is meant to be usable rather than comfortable.

Why comprehension and not agreement

Note what is absent from the three conditions. Nothing about the parties liking each other, sharing assumptions, or reaching the same conclusion. The canon has plenty to say elsewhere about how to argue — AN 2.42–51 on assemblies, and the discourses on how a question should be answered — and none of it is imported here.

The claim is narrower and stronger: the fuel of a discussion is comprehension, and if either side has it the talk continues. Two people who understand and disagree will keep going; two people who agree and understand nothing will stop.

Anyone who teaches has seen both cases. The second is the more dangerous, because agreement disguises it.

A note on duplicate discourses

Readers new to the Aṅguttara find these near-identical pairs suspicious, and it is worth explaining the practice rather than apologizing for it. The collection is organized numerically for recitation, and a set of three that fits two frames gets recorded under both, because the point of the arrangement is retrieval.

That is not the same as saying the pairs are redundant. AN 3.37 and AN 3.38, earlier in this collection, differ by one clause and the difference is the whole lesson. Here the difference is at both ends and the middle is untouched, which is a different pattern again: not one variable changed but the entire frame swapped around a fixed core.

A class can be shown the three patterns quickly — identical core with a changed reason, identical core with a changed frame, and genuine repetition — and after that the Aṅguttara stops looking careless.

Using it

The practical form is a single question to ask when a session dies: which of the three was I relying on? If the answer is that you assumed the second case — that the audience understood — the fix is not more energy but a check.

It is also a good discourse for a class that has begun to notice how repetitive the Aṅguttara is. Reading AN 3.43 and AN 3.44 back to back takes ninety seconds and teaches the reader how to read the whole collection: look at the ends, because the middle is often deliberately unchanged.

Key terms