On the especially nice fit between qualitative analysis and the known properties of conversation