Linearity of research design in ethnographic studies of speaking