Cultures without a written tradition depend entirely on the oral channel to transmit sometimes highly complex information. It is therefore not surprising that in the languages of such cultures linguistic devices evolve that enhance textual coherence, and thus comprehension. These devices should ideally also be economical in terms of morphosyntactic complexity in order to facilitate both production and comprehension. In this paper, I will argue that switch-attention (a term preferred over the traditional switch-reference) systems in temporal clauses fulfill these requirements of cohesion and complexity reduction, making them particularly apt for orally transmitting texts. Moreover, switch-reference systems seem to diffuse relatively easily. These features taken together are suggested to be (partly) responsible for the widely attested phenomenon in areas without a lengthy written tradition.